From 6c584fb68a6b5a8d008cc55167873d0624b1ce73 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E4=B8=80=E4=B8=AA=E6=88=90=E7=86=9F=E7=9A=84=E5=89=AA?= =?UTF-8?q?=E8=BE=91=E7=8C=BF?= <297668904+akahoz94@users.noreply.github.com> Date: Fri, 20 Nov 2026 03:43:07 +0800 Subject: [PATCH 1/3] feat(recording): pre-recording volume trims for microphone and computer audio Two sliders in the HUD device settings (-24..+12 dB, persisted in recording settings) trim each source before the take starts. The trim rides the native request on all three platforms and the browser fallback mix: - wgc-capture: AudioMixer takes a systemGain and applies it at mix time, sum-then-clamp like the microphone gain (queues stay unity). - pipewire-capture: system source reads audio.system.gain (already parsed, previously never sent by the Linux bridge). - ScreenCaptureKit: AudioTrackMixer scales the system track by the trim. - mixAudioTracks: GainNode trims for the browser path; mic-only and system-only unity stays context-free as before. The automatic 1.4x mic boost over system audio is unchanged; the user trim multiplies whichever base applies. Tests: audioMix (+3), app-settings (+1), C++ mixer system-gain exact-value case, request-shape and prefs literals updated; tsc both configs, biome, i18n:check clean; full vitest green except pre-existing release-script failures untouched by this change. --- electron/app-settings.test.ts | 21 ++++++ electron/app-settings.ts | 32 +++++++++ electron/ipc/handlers.ts | 14 +++- electron/ipc/recordingPrefs.test.ts | 2 + .../capture/linuxNativeCaptureSession.ts | 2 +- electron/native/pipewire-capture/src/main.rs | 4 +- .../AudioTrackMixer.swift | 6 +- .../ScreenCaptureRecorder.swift | 2 + .../wgc-capture/src/audio_sample_utils.cpp | 6 +- .../wgc-capture/src/audio_sample_utils.h | 2 + .../src/audio_sample_utils_test.cpp | 26 +++++--- electron/native/wgc-capture/src/main.cpp | 3 + .../ai-edition/v4/RecStage.test.tsx | 4 ++ .../launch/HudDeviceSettings.quality.test.tsx | 7 ++ src/components/launch/HudDeviceSettings.tsx | 65 ++++++++++++++++++- src/components/launch/LaunchWindow.module.css | 36 ++++++++++ src/components/launch/LaunchWindow.tsx | 34 ++++++++++ ...eenRecorder.nativeMacStartWarning.test.tsx | 2 +- src/hooks/useScreenRecorder.noCamera.test.tsx | 2 + .../useScreenRecorder.prefsRace.test.tsx | 2 + src/hooks/useScreenRecorder.ts | 28 ++++++-- src/i18n/locales/ar/launch.json | 2 + src/i18n/locales/cs/launch.json | 2 + src/i18n/locales/de/launch.json | 2 + src/i18n/locales/en/launch.json | 2 + src/i18n/locales/es/launch.json | 2 + src/i18n/locales/fr/launch.json | 2 + src/i18n/locales/it/launch.json | 2 + src/i18n/locales/ja-JP/launch.json | 2 + src/i18n/locales/ko-KR/launch.json | 2 + src/i18n/locales/nl/launch.json | 2 + src/i18n/locales/pl/launch.json | 2 + src/i18n/locales/pt-BR/launch.json | 2 + src/i18n/locales/ru/launch.json | 2 + src/i18n/locales/tr/launch.json | 2 + src/i18n/locales/vi/launch.json | 2 + src/i18n/locales/zh-CN/launch.json | 2 + src/i18n/locales/zh-TW/launch.json | 2 + src/lib/audioMix.test.ts | 44 +++++++++++++ src/lib/audioMix.ts | 45 +++++++++++-- src/lib/nativeLinuxRecording.ts | 2 + src/lib/nativeMacRecording.ts | 2 + src/lib/nativeWindowsRecording.ts | 2 + src/native/browserShim.ts | 4 ++ 44 files changed, 407 insertions(+), 26 deletions(-) diff --git a/electron/app-settings.test.ts b/electron/app-settings.test.ts index f3781734f8..36a1e72f86 100644 --- a/electron/app-settings.test.ts +++ b/electron/app-settings.test.ts @@ -62,6 +62,27 @@ describe("app settings store", () => { ); }); + it("persists the volume trims, reads missing keys as unity, and rejects out-of-range writes", () => { + const dir = temp(); + const file = path.join(dir, "recording-settings.json"); + const store = new AppSettingsStore(dir); + expect(store.getSnapshot().recording.micGainDb).toBe(0); + expect(store.getSnapshot().recording.systemAudioGainDb).toBe(0); + expect( + store.setRecordingPreferences({ micGainDb: -6, systemAudioGainDb: 3 }).recording, + ).toMatchObject({ micGainDb: -6, systemAudioGainDb: 3 }); + expect(new AppSettingsStore(dir).getSnapshot().recording).toMatchObject({ + micGainDb: -6, + systemAudioGainDb: 3, + }); + + writeFileSync(file, JSON.stringify({ micGainDb: 99, systemAudioGainDb: "loud" }), "utf8"); + expect(store.getSnapshot().recording.micGainDb).toBe(0); + expect(store.getSnapshot().recording.systemAudioGainDb).toBe(0); + expect(() => store.setRecordingPreferences({ micGainDb: 24 })).toThrow(TypeError); + expect(() => store.setRecordingPreferences({ systemAudioGainDb: -25 })).toThrow(TypeError); + }); + it("stores the last source beside the recording preferences", () => { const dir = temp(); const store = new AppSettingsStore(dir); diff --git a/electron/app-settings.ts b/electron/app-settings.ts index 81d334260c..6a7ce5c71c 100644 --- a/electron/app-settings.ts +++ b/electron/app-settings.ts @@ -6,6 +6,7 @@ import { type WebcamQualityId, webcamQualityFrom, } from "../src/hooks/webcamCaptureTarget"; +import { RECORDING_GAIN_DB_MAX, RECORDING_GAIN_DB_MIN } from "../src/lib/audioMix"; import type { CursorCaptureMode } from "../src/lib/recordingSession"; import { renameSyncWithFallback } from "./atomic-rename"; @@ -19,6 +20,10 @@ export interface RecordingPreferences { /** Capture resolution for the camera. See WEBCAM_QUALITY_PRESETS. */ camQuality: WebcamQualityId; systemAudioEnabled: boolean; + /** User trim on the microphone, dB relative to the automatic mix gain. */ + micGainDb: number; + /** User trim on the captured computer audio, dB. */ + systemAudioGainDb: number; cursorCaptureMode: CursorCaptureMode; /** Display captures on macOS and Windows. Opt-in: on Windows the icons also leave the real desktop while recording. */ hideDesktopIcons: boolean; @@ -41,6 +46,8 @@ export const DEFAULT_RECORDING_PREFERENCES: RecordingPreferences = { camDeviceName: null, camQuality: DEFAULT_WEBCAM_QUALITY, systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, @@ -101,6 +108,13 @@ function atomicWrite(userData: string, value: RawSettings): void { const bool = (value: unknown, fallback: boolean) => (typeof value === "boolean" ? value : fallback); const nullableString = (value: unknown, fallback: string | null) => value === null || typeof value === "string" ? value : fallback; +const gainDb = (value: unknown, fallback: number) => + typeof value === "number" && + Number.isFinite(value) && + value >= RECORDING_GAIN_DB_MIN && + value <= RECORDING_GAIN_DB_MAX + ? value + : fallback; function parseRecording(raw: RawSettings): RecordingPreferences { return { @@ -117,6 +131,13 @@ function parseRecording(raw: RawSettings): RecordingPreferences { raw.systemAudioEnabled, DEFAULT_RECORDING_PREFERENCES.systemAudioEnabled, ), + // Absent from settings files written before the volume trims existed; + // unity (0 dB) is what every such installation recorded at. + micGainDb: gainDb(raw.micGainDb, DEFAULT_RECORDING_PREFERENCES.micGainDb), + systemAudioGainDb: gainDb( + raw.systemAudioGainDb, + DEFAULT_RECORDING_PREFERENCES.systemAudioGainDb, + ), cursorCaptureMode: raw.cursorCaptureMode === "system" || raw.cursorCaptureMode === "editable-overlay" ? raw.cursorCaptureMode @@ -186,6 +207,17 @@ function validateRecordingPatch(patch: Partial): void { if (key === "camQuality" && !WEBCAM_QUALITY_IDS.includes(value as WebcamQualityId)) { throw new TypeError("camQuality is invalid"); } + if ( + (key === "micGainDb" || key === "systemAudioGainDb") && + (typeof value !== "number" || + !Number.isFinite(value) || + value < RECORDING_GAIN_DB_MIN || + value > RECORDING_GAIN_DB_MAX) + ) { + throw new TypeError( + `${key} must be a number in [${RECORDING_GAIN_DB_MIN}, ${RECORDING_GAIN_DB_MAX}]`, + ); + } } } diff --git a/electron/ipc/handlers.ts b/electron/ipc/handlers.ts index e4bab0f5cb..a6668a1d18 100644 --- a/electron/ipc/handlers.ts +++ b/electron/ipc/handlers.ts @@ -736,6 +736,10 @@ export interface RecordingPrefs { /** Capture resolution for the camera. See WEBCAM_QUALITY_PRESETS. */ camQuality: WebcamQualityId; systemAudioEnabled: boolean; + /** User trim on the microphone, dB relative to the automatic mix gain. */ + micGainDb: number; + /** User trim on the captured computer audio, dB. */ + systemAudioGainDb: number; cursorCaptureMode: CursorCaptureMode; hideDesktopIcons: boolean; /** Whether a fresh take gets automatic zooms on import. Persisted; defaults on. */ @@ -750,6 +754,8 @@ const defaultRecordingPrefs: RecordingPrefs = { camDeviceName: null, camQuality: DEFAULT_WEBCAM_QUALITY, systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, @@ -2795,7 +2801,7 @@ export function registerIpcHandlers( fps: request.video.fps, ...(request.video.bitrate ? { bitrate: request.video.bitrate } : {}), audio: { - system: { enabled: request.audio.system.enabled }, + system: { enabled: request.audio.system.enabled, gain: request.audio.system.gain ?? 1 }, microphone: { enabled: request.audio.microphone.enabled, ...(request.audio.microphone.deviceName @@ -2899,7 +2905,10 @@ export function registerIpcHandlers( fps: request.video.fps, ...(request.video.bitrate ? { bitrate: request.video.bitrate } : {}), audio: { - system: { enabled: request.audio.system.enabled }, + system: { + enabled: request.audio.system.enabled, + gain: request.audio.system.gain ?? 1, + }, microphone: { enabled: request.audio.microphone.enabled, ...(request.audio.microphone.deviceName @@ -3190,6 +3199,7 @@ export function registerIpcHandlers( microphoneDeviceId: request.audio.microphone.deviceId ?? null, microphoneDeviceName: request.audio.microphone.deviceName ?? null, microphoneGain: request.audio.microphone.gain, + systemGain: request.audio.system.gain ?? 1, webcamEnabled: request.webcam.enabled, webcamDeviceId: request.webcam.deviceId ?? null, webcamDeviceName: request.webcam.deviceName ?? null, diff --git a/electron/ipc/recordingPrefs.test.ts b/electron/ipc/recordingPrefs.test.ts index 2c3317af93..7a65bb7d4b 100644 --- a/electron/ipc/recordingPrefs.test.ts +++ b/electron/ipc/recordingPrefs.test.ts @@ -21,6 +21,8 @@ const defaults: RecordingPrefs = { camDeviceName: null, camQuality: "2160p", systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, diff --git a/electron/native-bridge/capture/linuxNativeCaptureSession.ts b/electron/native-bridge/capture/linuxNativeCaptureSession.ts index fa4a3767dc..e6f1e4ca86 100644 --- a/electron/native-bridge/capture/linuxNativeCaptureSession.ts +++ b/electron/native-bridge/capture/linuxNativeCaptureSession.ts @@ -72,7 +72,7 @@ export interface LinuxCaptureConfig { fps: number; bitrate?: number; audio: { - system: { enabled: boolean }; + system: { enabled: boolean; gain?: number }; microphone: { enabled: boolean; deviceName?: string; gain: number }; }; maxCursorSamples: number; diff --git a/electron/native/pipewire-capture/src/main.rs b/electron/native/pipewire-capture/src/main.rs index 8d99997e68..732139dc15 100644 --- a/electron/native/pipewire-capture/src/main.rs +++ b/electron/native/pipewire-capture/src/main.rs @@ -79,6 +79,8 @@ struct AudioRequest { #[serde(rename_all = "camelCase", default)] struct SystemAudioRequest { enabled: bool, + /// Linear multiplier the UI applies to computer-audio level. + gain: Option, } #[derive(Debug, Default, Deserialize)] @@ -199,7 +201,7 @@ impl Request { // The default SINK's monitor: what is being played, not what a // microphone hears. capture_sink: true, - gain: 1.0, + gain: audio.system.gain.filter(|gain| *gain > 0.0).unwrap_or(1.0), bitrate: DEFAULT_AUDIO_BITRATE, }); } diff --git a/electron/native/screencapturekit/Sources/OpenScreenCaptureCore/AudioTrackMixer.swift b/electron/native/screencapturekit/Sources/OpenScreenCaptureCore/AudioTrackMixer.swift index e0f3b17852..32488ddf0c 100644 --- a/electron/native/screencapturekit/Sources/OpenScreenCaptureCore/AudioTrackMixer.swift +++ b/electron/native/screencapturekit/Sources/OpenScreenCaptureCore/AudioTrackMixer.swift @@ -106,6 +106,7 @@ public final class AudioTrackMixer { private let includesSystemAudio: Bool private let includesMicrophone: Bool private let microphoneGain: Float + private let systemGain: Float private let outputFormatDescription: CMAudioFormatDescription? private var sources = [SourceTimeline](repeating: SourceTimeline(), count: Source.allCases.count) @@ -136,6 +137,7 @@ public final class AudioTrackMixer { includesSystemAudio: Bool, includesMicrophone: Bool, microphoneGain: Double, + systemGain: Double = 1, clock: @escaping TimelineClock ) { self.input = input @@ -146,6 +148,8 @@ public final class AudioTrackMixer { // does this. A non-finite or negative value would poison every mixed sample. let sanitized = microphoneGain.isFinite ? max(0, microphoneGain) : 1 self.microphoneGain = Float(sanitized) + let sanitizedSystem = systemGain.isFinite ? max(0, systemGain) : 1 + self.systemGain = Float(sanitizedSystem) self.outputFormatDescription = Self.makeOutputFormatDescription() } @@ -397,7 +401,7 @@ public final class AudioTrackMixer { private func gain(for source: Source) -> Float { switch source { case .system: - return 1 + return systemGain case .microphone: return microphoneGain } diff --git a/electron/native/screencapturekit/Sources/OpenScreenScreenCaptureKitHelper/ScreenCaptureRecorder.swift b/electron/native/screencapturekit/Sources/OpenScreenScreenCaptureKitHelper/ScreenCaptureRecorder.swift index 3b1cce0c7d..d17479246c 100644 --- a/electron/native/screencapturekit/Sources/OpenScreenScreenCaptureKitHelper/ScreenCaptureRecorder.swift +++ b/electron/native/screencapturekit/Sources/OpenScreenScreenCaptureKitHelper/ScreenCaptureRecorder.swift @@ -32,6 +32,7 @@ struct RecordingRequest: Decodable { struct Audio: Decodable { struct SystemAudio: Decodable { let enabled: Bool + let gain: Double? } struct Microphone: Decodable { @@ -973,6 +974,7 @@ final class ScreenCaptureRecorder: NSObject, SCStreamOutput, SCStreamDelegate { includesSystemAudio: request.audio.system.enabled, includesMicrophone: nativeMicrophoneEnabled, microphoneGain: request.audio.microphone.gain, + systemGain: request.audio.system.gain ?? 1, clock: { [weak self] in self?.timelineNow() ?? .invalid } ) } diff --git a/electron/native/wgc-capture/src/audio_sample_utils.cpp b/electron/native/wgc-capture/src/audio_sample_utils.cpp index 07d35a162f..8964a0c5c2 100644 --- a/electron/native/wgc-capture/src/audio_sample_utils.cpp +++ b/electron/native/wgc-capture/src/audio_sample_utils.cpp @@ -898,6 +898,7 @@ AudioMixer::AudioMixer( bool includeSystem, bool includeMicrophone, double microphoneGain, + double systemGain, OutputCallback output) : format_(format), systemFormat_(systemFormat), @@ -905,6 +906,7 @@ AudioMixer::AudioMixer( includeSystem_(includeSystem), includeMicrophone_(includeMicrophone), microphoneGain_(microphoneGain), + systemGain_(systemGain), output_(std::move(output)) {} AudioMixer::~AudioMixer() { @@ -1146,7 +1148,9 @@ void AudioMixer::mixLoop() { mixedChunk.assign(chunkBytes, 0); if (includeSystem_) { pop(systemQueue_, systemStarved_, sourceChunk, chunkBytes); - mixAudioInPlace(mixedChunk, sourceChunk.data(), static_cast(sourceChunk.size()), format_); + mixAudioInPlace( + mixedChunk, sourceChunk.data(), static_cast(sourceChunk.size()), format_, + systemGain_); } if (includeMicrophone_) { pop(microphoneQueue_, microphoneStarved_, sourceChunk, chunkBytes); diff --git a/electron/native/wgc-capture/src/audio_sample_utils.h b/electron/native/wgc-capture/src/audio_sample_utils.h index b1dafec4c8..18d784cc48 100644 --- a/electron/native/wgc-capture/src/audio_sample_utils.h +++ b/electron/native/wgc-capture/src/audio_sample_utils.h @@ -230,6 +230,7 @@ class AudioMixer { bool includeSystem, bool includeMicrophone, double microphoneGain, + double systemGain, OutputCallback output); ~AudioMixer(); @@ -262,6 +263,7 @@ class AudioMixer { bool includeSystem_ = false; bool includeMicrophone_ = false; double microphoneGain_ = 1.0; + double systemGain_ = 1.0; OutputCallback output_; std::mutex mutex_; std::condition_variable cv_; diff --git a/electron/native/wgc-capture/src/audio_sample_utils_test.cpp b/electron/native/wgc-capture/src/audio_sample_utils_test.cpp index 3654e5a78d..1d98eb1559 100644 --- a/electron/native/wgc-capture/src/audio_sample_utils_test.cpp +++ b/electron/native/wgc-capture/src/audio_sample_utils_test.cpp @@ -1290,7 +1290,7 @@ void runResampleContinuityFixedTests() { const AudioInputFormat sysFmt = makeFormat(MFAudioFormat_Float, 44100, 2, 32); const AudioInputFormat micFmt = makeFormat(MFAudioFormat_Float, 48000, 2, 32); AudioMixer mixer( - mixFmt, sysFmt, micFmt, true, true, 1.0, + mixFmt, sysFmt, micFmt, true, true, 1.0, 1.0, [](const BYTE*, DWORD, int64_t, int64_t) { return true; }); const std::vector sysSignal = stereoFloatDualSine(44100, 997.0, 501.0, 0.5, 4321); @@ -1466,7 +1466,7 @@ void runResampleContinuityFixedTests() { const AudioInputFormat mixFmt = makeFormat(MFAudioFormat_PCM, 44100, 2, 16); const AudioInputFormat sysFmt = makeFormat(MFAudioFormat_Float, 48000, 2, 32); AudioMixer mixer( - mixFmt, sysFmt, sysFmt, true, false, 1.0, + mixFmt, sysFmt, sysFmt, true, false, 1.0, 1.0, [](const BYTE*, DWORD, int64_t, int64_t) { return true; }); std::vector preGap(150 * sysFmt.blockAlign, 0); @@ -1509,7 +1509,7 @@ void runResampleContinuityFixedTests() { const AudioInputFormat intMixFmt = makeFormat(MFAudioFormat_PCM, 48000, 2, 16); const AudioInputFormat intSrcFmt = makeFormat(MFAudioFormat_Float, 96000, 2, 32); AudioMixer intMixer( - intMixFmt, intSrcFmt, intSrcFmt, true, false, 1.0, + intMixFmt, intSrcFmt, intSrcFmt, true, false, 1.0, 1.0, [](const BYTE*, DWORD, int64_t, int64_t) { return true; }); AudioDecimatorState intContinuing; std::vector intPiece; @@ -1541,7 +1541,7 @@ void runResampleContinuityFixedTests() { const AudioInputFormat sameMixFmt = makeFormat(MFAudioFormat_PCM, 48000, 2, 16); const AudioInputFormat sameSrcFmt = makeFormat(MFAudioFormat_Float, 48000, 2, 32); AudioMixer sameMixer( - sameMixFmt, sameSrcFmt, sameSrcFmt, true, false, 1.0, + sameMixFmt, sameSrcFmt, sameSrcFmt, true, false, 1.0, 1.0, [](const BYTE*, DWORD, int64_t, int64_t) { return true; }); sameMixer.pushSystem(preGap.data(), static_cast(preGap.size())); AudioMixerTestAccess::setSystemStarved(sameMixer, true); @@ -2643,7 +2643,7 @@ int main(int argc, char** argv) { size_t zeroChunks = 0; AudioMixer mixer( - target48k, system96k, mic48k, true, includeMic, 1.0, + target48k, system96k, mic48k, true, includeMic, 1.0, 1.0, [&](const BYTE* data, DWORD byteCount, int64_t, int64_t) { std::scoped_lock lock(guard); chunkCount += 1; @@ -2780,11 +2780,11 @@ int main(int argc, char** argv) { return bytes; }; const auto steadySample = [&](bool includeSystem, double system, double mic, - double gain) -> int { + double gain, double systemGain = 1.0) -> int { std::mutex guard; std::vector collected; AudioMixer mixer( - target48k, f32stereo, f32stereo, includeSystem, true, gain, + target48k, f32stereo, f32stereo, includeSystem, true, gain, systemGain, [&](const BYTE* data, DWORD byteCount, int64_t, int64_t) { std::scoped_lock lock(guard); collected.insert(collected.end(), data, data + byteCount); @@ -2850,6 +2850,16 @@ int main(int argc, char** argv) { std::cout << "GAIN_RAW unity " << detail << std::endl; expect("mixer-unity-gain-passthrough", std::abs(got - 29490) <= 2, detail); } + // (4) The system stream carries its own gain: 0.8 * 0.5 + 0.3 = 0.7 + // -> 22937. A mixer that ignored systemGain would read 0.8 + 0.3 = 1.1 + // clamped at the rail instead. + { + const int got = steadySample(true, 0.8, 0.3, 1.0, 0.5); + char detail[96]{}; + sprintf_s(detail, "got=%d want=22937", got); + std::cout << "GAIN_RAW system-gain " << detail << std::endl; + expect("mixer-system-gain-applied-at-mix", std::abs(got - 22937) <= 2, detail); + } } // --- Capture jitter: the mixer's cushion (getopenscreen/openscreen#911) --- @@ -2897,7 +2907,7 @@ int main(int argc, char** argv) { std::mutex guard; std::vector left; AudioMixer mixer( - target48k, f32, f32, includeSystem, includeMic, 1.0, + target48k, f32, f32, includeSystem, includeMic, 1.0, 1.0, [&](const BYTE* data, DWORD byteCount, int64_t, int64_t) { std::scoped_lock lock(guard); const auto* samples = reinterpret_cast(data); diff --git a/electron/native/wgc-capture/src/main.cpp b/electron/native/wgc-capture/src/main.cpp index fe38c8a17d..1dc78156a6 100644 --- a/electron/native/wgc-capture/src/main.cpp +++ b/electron/native/wgc-capture/src/main.cpp @@ -58,6 +58,7 @@ struct CaptureConfig { std::string microphoneDeviceId; std::string microphoneDeviceName; double microphoneGain = 1.0; + double systemGain = 1.0; std::string webcamDeviceId; std::string webcamDeviceName; std::string webcamDirectShowClsid; @@ -592,6 +593,7 @@ bool parseConfig(const std::string& json, CaptureConfig& config) { config.microphoneDeviceId = findString(json, "microphoneDeviceId"); config.microphoneDeviceName = findString(json, "microphoneDeviceName"); config.microphoneGain = findDouble(json, "microphoneGain", 1.0); + config.systemGain = findDouble(json, "systemGain", 1.0); config.webcamDeviceId = findString(json, "webcamDeviceId"); config.webcamDeviceName = findString(json, "webcamDeviceName"); config.webcamDirectShowClsid = findString(json, "webcamDirectShowClsid"); @@ -1424,6 +1426,7 @@ int wmain(int argc, wchar_t* argv[]) { config.captureSystemAudio, config.captureMic, config.microphoneGain, + config.systemGain, [&](const BYTE* data, DWORD byteCount, int64_t timestampHns, int64_t durationHns) { if (!encoder.writeAudio(data, byteCount, timestampHns, durationHns)) { encodeFailed = true; diff --git a/src/components/ai-edition/v4/RecStage.test.tsx b/src/components/ai-edition/v4/RecStage.test.tsx index 49230e2784..be13f65db0 100644 --- a/src/components/ai-edition/v4/RecStage.test.tsx +++ b/src/components/ai-edition/v4/RecStage.test.tsx @@ -420,6 +420,8 @@ describe("RecStage controls", () => { camDeviceName: null, camQuality: "2160p", systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, @@ -632,6 +634,8 @@ describe("RecStage controls", () => { camDeviceName: null, camQuality: "2160p", systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, diff --git a/src/components/launch/HudDeviceSettings.quality.test.tsx b/src/components/launch/HudDeviceSettings.quality.test.tsx index 126ce93148..00e3e167ea 100644 --- a/src/components/launch/HudDeviceSettings.quality.test.tsx +++ b/src/components/launch/HudDeviceSettings.quality.test.tsx @@ -19,6 +19,9 @@ const labels: HudDeviceSettingsLabels = { camera: "Camera", micLevel: "Input level", micHint: "Speak to check", + micGain: "Microphone volume", + systemAudio: "System audio", + systemGain: "Computer volume", noMicrophones: "No microphone found", searching: "Searching...", noCameras: "No camera found", @@ -55,6 +58,10 @@ function renderPanel(overrides: Partial[0]> checkingForUpdates={false} cameraQuality="1440p" onSelectCameraQuality={onSelectCameraQuality} + micGainDb={0} + systemGainDb={0} + onMicGainChange={vi.fn()} + onSystemGainChange={vi.fn()} onSelectMic={vi.fn()} onSelectCamera={vi.fn()} onCheckForUpdates={vi.fn()} diff --git a/src/components/launch/HudDeviceSettings.tsx b/src/components/launch/HudDeviceSettings.tsx index 164b2f370a..84343c4940 100644 --- a/src/components/launch/HudDeviceSettings.tsx +++ b/src/components/launch/HudDeviceSettings.tsx @@ -1,10 +1,11 @@ import { Check, X } from "lucide-react"; -import { memo, useEffect, useRef } from "react"; +import { type CSSProperties, memo, useEffect, useRef } from "react"; import { useAudioLevelMeter } from "../../hooks/useAudioLevelMeter"; import type { CameraDevice } from "../../hooks/useCameraDevices"; import { useCameraPreviewStream } from "../../hooks/useCameraPreviewStream"; import type { MicrophoneDevice } from "../../hooks/useMicrophoneDevices"; import { WEBCAM_QUALITY_IDS, type WebcamQualityId } from "../../hooks/webcamCaptureTarget"; +import { RECORDING_GAIN_DB_MAX, RECORDING_GAIN_DB_MIN } from "../../lib/audioMix"; import { Tooltip } from "../ui/tooltip"; import styles from "./LaunchWindow.module.css"; @@ -18,6 +19,9 @@ export interface HudDeviceSettingsLabels { camera: string; micLevel: string; micHint: string; + micGain: string; + systemAudio: string; + systemGain: string; noMicrophones: string; searching: string; noCameras: string; @@ -46,6 +50,43 @@ const LevelMeter = memo(function LevelMeter({ level }: { level: number }) { ); }); +/** Pre-recording volume trim, in dB. The value rides the recording request; + * it does not retune a take already in flight. */ +function GainSlider({ + label, + value, + onChange, + testId, +}: { + label: string; + value: number; + onChange: (db: number) => void; + testId: string; +}) { + return ( +
+ {label} + onChange(Number(event.target.value))} + style={ + { + "--slider-pct": `${((value - RECORDING_GAIN_DB_MIN) / (RECORDING_GAIN_DB_MAX - RECORDING_GAIN_DB_MIN)) * 100}%`, + } as CSSProperties + } + className={styles.hudModalGainSlider} + /> + {value > 0 ? `+${value}` : value} dB +
+ ); +} + function CameraPreview({ stream, error, @@ -102,6 +143,10 @@ export const HudDeviceSettings = memo(function HudDeviceSettings({ checkingForUpdates, cameraQuality, onSelectCameraQuality, + micGainDb, + systemGainDb, + onMicGainChange, + onSystemGainChange, onSelectMic, onSelectCamera, onCheckForUpdates, @@ -124,6 +169,10 @@ export const HudDeviceSettings = memo(function HudDeviceSettings({ checkingForUpdates: boolean; cameraQuality: WebcamQualityId; onSelectCameraQuality: (quality: WebcamQualityId) => void; + micGainDb: number; + systemGainDb: number; + onMicGainChange: (db: number) => void; + onSystemGainChange: (db: number) => void; onSelectMic: (device: MicrophoneDevice) => void; onSelectCamera: (device: CameraDevice) => void; onCheckForUpdates: () => void; @@ -195,9 +244,23 @@ export const HudDeviceSettings = memo(function HudDeviceSettings({
{labels.micHint}
+ ) : null} +
{labels.systemAudio}
+ +
{labels.camera}
{cameraLoading ? (
{labels.searching}
diff --git a/src/components/launch/LaunchWindow.module.css b/src/components/launch/LaunchWindow.module.css index 0f522b32ef..4837b762cc 100644 --- a/src/components/launch/LaunchWindow.module.css +++ b/src/components/launch/LaunchWindow.module.css @@ -211,6 +211,42 @@ color: rgba(255, 255, 255, 0.6); } +.hudModalGainRow { + display: flex; + align-items: center; + gap: 8px; + padding: 6px 10px; +} + +.hudModalGainSlider { + flex: 1; + min-width: 0; + height: 6px; + border-radius: 999px; + background: linear-gradient( + to right, + var(--brand) 0%, + var(--brand) var(--slider-pct, 0%), + var(--border-hi) var(--slider-pct, 0%), + var(--border-hi) 100% + ); + outline: none; + cursor: pointer; +} + +.hudModalGainSlider:focus-visible::-webkit-slider-thumb { + box-shadow: 0 2px 8px rgba(255, 255, 255, 0.32), 0 0 0 4px rgba(52, 178, 123, 0.35); +} + +.hudModalGainValue { + font-size: 12px; + color: rgba(255, 255, 255, 0.6); + white-space: nowrap; + min-width: 44px; + text-align: right; + font-variant-numeric: tabular-nums; +} + /* About block at the foot of the settings panel: version on the left, the update check on the right. Same 10px gutter as the meter rows above it so the panel keeps one left edge. */ .hudModalAboutRow { diff --git a/src/components/launch/LaunchWindow.tsx b/src/components/launch/LaunchWindow.tsx index 754b1adfac..02eb11c8f8 100644 --- a/src/components/launch/LaunchWindow.tsx +++ b/src/components/launch/LaunchWindow.tsx @@ -139,6 +139,10 @@ export function LaunchWindow() { setMicrophoneDeviceName, systemAudioEnabled, setSystemAudioEnabled, + micGainDb, + setMicGainDb, + systemAudioGainDb, + setSystemAudioGainDb, webcamEnabled, setWebcamEnabled, webcamDeviceId, @@ -974,6 +978,8 @@ export function LaunchWindow() { micDeviceId?: string; micDeviceName?: string; systemAudioEnabled?: boolean; + micGainDb?: number; + systemAudioGainDb?: number; cursorCaptureMode?: "editable-overlay" | "system"; }) => { void window.electronAPI?.setRecordingPrefs?.(patch).catch((error) => { @@ -990,6 +996,27 @@ export function LaunchWindow() { persistRecordingPrefs({ systemAudioEnabled: next }); }, [controlsLocked, persistRecordingPrefs, setSystemAudioEnabled, systemAudioEnabled]); + // Same guard as the camera quality picker: the gear is disabled mid-take but + // an open panel stays mounted. The trim rides the NEXT take's request, so a + // change here never retunes the one in flight. + const handleMicGainChange = useCallback( + (db: number) => { + if (controlsLocked) return; + setMicGainDb(db); + persistRecordingPrefs({ micGainDb: db }); + }, + [controlsLocked, persistRecordingPrefs, setMicGainDb], + ); + + const handleSystemGainChange = useCallback( + (db: number) => { + if (controlsLocked) return; + setSystemAudioGainDb(db); + persistRecordingPrefs({ systemAudioGainDb: db }); + }, + [controlsLocked, persistRecordingPrefs, setSystemAudioGainDb], + ); + const toggleCursorMode = useCallback(() => { if (controlsLocked) return; const next = cursorCaptureMode === "editable-overlay" ? "system" : "editable-overlay"; @@ -1206,6 +1233,9 @@ export function LaunchWindow() { camera: t("webcam.cameraDevice"), micLevel: t("deviceSettings.micLevel"), micHint: t("deviceSettings.micHint"), + micGain: t("deviceSettings.micGain"), + systemAudio: t("audio.systemAudio"), + systemGain: t("deviceSettings.systemGain"), noMicrophones: t("deviceSettings.noMicrophones"), searching: t("webcam.searching"), noCameras: t("webcam.noneFound"), @@ -1467,6 +1497,10 @@ export function LaunchWindow() { checkingForUpdates={isCheckingForUpdates} cameraQuality={webcamQuality} onSelectCameraQuality={handleSelectCameraQuality} + micGainDb={micGainDb} + systemGainDb={systemAudioGainDb} + onMicGainChange={handleMicGainChange} + onSystemGainChange={handleSystemGainChange} onSelectMic={handleSelectMicDevice} onSelectCamera={handleSelectCameraDevice} onCheckForUpdates={handleCheckForUpdates} diff --git a/src/hooks/useScreenRecorder.nativeMacStartWarning.test.tsx b/src/hooks/useScreenRecorder.nativeMacStartWarning.test.tsx index 4e4009e803..337a3524ff 100644 --- a/src/hooks/useScreenRecorder.nativeMacStartWarning.test.tsx +++ b/src/hooks/useScreenRecorder.nativeMacStartWarning.test.tsx @@ -102,7 +102,7 @@ describe("useScreenRecorder native macOS start warnings", () => { expect(api.startNativeMacRecording).toHaveBeenCalledWith( expect.objectContaining({ audio: { - system: { enabled: false }, + system: { enabled: false, gain: 1 }, microphone: expect.objectContaining({ enabled: true, deviceId: "chromium-device-id", diff --git a/src/hooks/useScreenRecorder.noCamera.test.tsx b/src/hooks/useScreenRecorder.noCamera.test.tsx index 4958a8fbf1..b0f7582b9e 100644 --- a/src/hooks/useScreenRecorder.noCamera.test.tsx +++ b/src/hooks/useScreenRecorder.noCamera.test.tsx @@ -33,6 +33,8 @@ function prefs(camEnabled: boolean): RecordingPrefs { camDeviceName: null, camQuality: "2160p", systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, diff --git a/src/hooks/useScreenRecorder.prefsRace.test.tsx b/src/hooks/useScreenRecorder.prefsRace.test.tsx index ae84c2cd71..1c22681322 100644 --- a/src/hooks/useScreenRecorder.prefsRace.test.tsx +++ b/src/hooks/useScreenRecorder.prefsRace.test.tsx @@ -27,6 +27,8 @@ function prefs(micEnabled: boolean): RecordingPrefs { camDeviceName: null, camQuality: "2160p", systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", hideDesktopIcons: false, autoZoomEnabled: true, diff --git a/src/hooks/useScreenRecorder.ts b/src/hooks/useScreenRecorder.ts index f4ef513918..19e0630407 100644 --- a/src/hooks/useScreenRecorder.ts +++ b/src/hooks/useScreenRecorder.ts @@ -2,7 +2,7 @@ import { fixWebmDuration } from "@fix-webm-duration/fix"; import { useCallback, useEffect, useRef, useState } from "react"; import { toast } from "sonner"; import { useScopedT } from "@/contexts/I18nContext"; -import { mixAudioTracks, nativeMicrophoneGain } from "@/lib/audioMix"; +import { mixAudioTracks, nativeMicrophoneGain, recordingGainScalar } from "@/lib/audioMix"; import { canTurnCameraOn } from "@/lib/cameraAvailability"; import { shouldRecordMicrophone } from "@/lib/microphoneAvailability"; import { @@ -115,6 +115,10 @@ type UseScreenRecorderReturn = { setWebcamDeviceName: (deviceName: string | undefined) => void; systemAudioEnabled: boolean; setSystemAudioEnabled: (enabled: boolean) => void; + micGainDb: number; + setMicGainDb: (db: number) => void; + systemAudioGainDb: number; + setSystemAudioGainDb: (db: number) => void; webcamEnabled: boolean; setWebcamEnabled: (enabled: boolean) => Promise; cursorCaptureMode: CursorCaptureMode; @@ -309,6 +313,8 @@ export function useScreenRecorder(): UseScreenRecorderReturn { const [webcamQuality, setWebcamQuality] = useState(DEFAULT_WEBCAM_QUALITY); const [webcamDeviceName, setWebcamDeviceName] = useState(undefined); const [systemAudioEnabled, setSystemAudioEnabled] = useState(false); + const [micGainDb, setMicGainDb] = useState(0); + const [systemAudioGainDb, setSystemAudioGainDb] = useState(0); const [webcamEnabled, setWebcamEnabledState] = useState(false); const [cursorCaptureMode, setCursorCaptureMode] = useState("editable-overlay"); const [softwareEncoderFallbackNoticeVisible, setSoftwareEncoderFallbackNoticeVisible] = @@ -337,6 +343,8 @@ export function useScreenRecorder(): UseScreenRecorderReturn { camDeviceName?: string | null; camQuality?: WebcamQualityId | null; systemAudioEnabled: boolean; + micGainDb?: number; + systemAudioGainDb?: number; cursorCaptureMode: CursorCaptureMode; }) => { if (cancelled) return; @@ -370,6 +378,8 @@ export function useScreenRecorder(): UseScreenRecorderReturn { } setWebcamQuality(webcamQualityFrom(prefs.camQuality)); setSystemAudioEnabled(prefs.systemAudioEnabled); + setMicGainDb(prefs.micGainDb ?? 0); + setSystemAudioGainDb(prefs.systemAudioGainDb ?? 0); setCursorCaptureMode(prefs.cursorCaptureMode); setRecordingPrefsLoaded(true); }; @@ -1352,13 +1362,14 @@ export function useScreenRecorder(): UseScreenRecorderReturn { audio: { system: { enabled: systemAudioEnabled, + gain: recordingGainScalar(systemAudioGainDb), }, microphone: { enabled: recordMicrophone, deviceId: microphoneDeviceId, deviceName: microphoneDeviceName, // Boosted only when the mic has to sit over system audio. - gain: nativeMicrophoneGain(systemAudioEnabled), + gain: nativeMicrophoneGain(systemAudioEnabled) * recordingGainScalar(micGainDb), }, }, webcam: { @@ -1519,13 +1530,14 @@ export function useScreenRecorder(): UseScreenRecorderReturn { audio: { system: { enabled: systemAudioEnabled, + gain: recordingGainScalar(systemAudioGainDb), }, microphone: { enabled: recordMicrophone, deviceId: microphoneDeviceId, deviceName: microphoneDeviceName, // Boosted only over system audio, like the Windows request above. - gain: nativeMicrophoneGain(systemAudioEnabled), + gain: nativeMicrophoneGain(systemAudioEnabled) * recordingGainScalar(micGainDb), }, }, webcam: { @@ -1641,7 +1653,7 @@ export function useScreenRecorder(): UseScreenRecorderReturn { fps: TARGET_FRAME_RATE, }, audio: { - system: { enabled: systemAudioEnabled }, + system: { enabled: systemAudioEnabled, gain: recordingGainScalar(systemAudioGainDb) }, microphone: { enabled: recordMicrophone, // The device LABEL, not the id. Chromium's deviceId is an opaque @@ -1653,7 +1665,7 @@ export function useScreenRecorder(): UseScreenRecorderReturn { // helper then fell back to the session default source. ...(microphoneDeviceName ? { deviceName: microphoneDeviceName } : {}), // Boosted only over system audio, like the Windows request above. - gain: nativeMicrophoneGain(systemAudioEnabled), + gain: nativeMicrophoneGain(systemAudioEnabled) * recordingGainScalar(micGainDb), }, }, cursor: { mode: cursorCaptureMode }, @@ -2179,6 +2191,8 @@ export function useScreenRecorder(): UseScreenRecorderReturn { const { context: mixingCtx, track: mixedTrack } = mixAudioTracks({ systemAudioTrack, micAudioTrack, + micGainDb, + systemGainDb: systemAudioGainDb, }); if (mixingCtx) { mixingContext.current = mixingCtx; @@ -2704,6 +2718,10 @@ export function useScreenRecorder(): UseScreenRecorderReturn { setWebcamDeviceName, systemAudioEnabled, setSystemAudioEnabled, + micGainDb, + setMicGainDb, + systemAudioGainDb, + setSystemAudioGainDb, webcamEnabled, setWebcamEnabled, cursorCaptureMode, diff --git a/src/i18n/locales/ar/launch.json b/src/i18n/locales/ar/launch.json index 538b34e262..de299ba912 100644 --- a/src/i18n/locales/ar/launch.json +++ b/src/i18n/locales/ar/launch.json @@ -99,6 +99,8 @@ "close": "إغلاق", "micLevel": "مستوى الإدخال", "micHint": "تحدّث لاختبار الميكروفون", + "micGain": "مستوى صوت الميكروفون", + "systemGain": "مستوى صوت الكمبيوتر", "noMicrophones": "لم يتم العثور على ميكروفون", "preview": "معاينة", "previewUnavailable": "المعاينة غير متاحة", diff --git a/src/i18n/locales/cs/launch.json b/src/i18n/locales/cs/launch.json index a5a63939a2..a85b8b4a67 100644 --- a/src/i18n/locales/cs/launch.json +++ b/src/i18n/locales/cs/launch.json @@ -99,6 +99,8 @@ "close": "Zavřít", "micLevel": "Úroveň vstupu", "micHint": "Mluvte pro kontrolu mikrofonu", + "micGain": "Hlasitost mikrofonu", + "systemGain": "Hlasitost zvuku počítače", "noMicrophones": "Mikrofon nenalezen", "preview": "Náhled", "previewUnavailable": "Náhled nedostupný", diff --git a/src/i18n/locales/de/launch.json b/src/i18n/locales/de/launch.json index 204b6e8589..895eb0c953 100644 --- a/src/i18n/locales/de/launch.json +++ b/src/i18n/locales/de/launch.json @@ -99,6 +99,8 @@ "close": "Schließen", "micLevel": "Eingangspegel", "micHint": "Sprich, um dein Mikrofon zu prüfen", + "micGain": "Mikrofonlautstärke", + "systemGain": "Computerlautstärke", "noMicrophones": "Kein Mikrofon gefunden", "preview": "Vorschau", "previewUnavailable": "Vorschau nicht verfügbar", diff --git a/src/i18n/locales/en/launch.json b/src/i18n/locales/en/launch.json index 4987f2396e..4ccc5e06ec 100644 --- a/src/i18n/locales/en/launch.json +++ b/src/i18n/locales/en/launch.json @@ -99,6 +99,8 @@ "close": "Close", "micLevel": "Input level", "micHint": "Speak to check your microphone", + "micGain": "Microphone volume", + "systemGain": "Computer volume", "noMicrophones": "No microphone found", "preview": "Preview", "previewUnavailable": "Preview unavailable", diff --git a/src/i18n/locales/es/launch.json b/src/i18n/locales/es/launch.json index fc230bba7f..fd0c35c317 100644 --- a/src/i18n/locales/es/launch.json +++ b/src/i18n/locales/es/launch.json @@ -99,6 +99,8 @@ "close": "Cerrar", "micLevel": "Nivel de entrada", "micHint": "Habla para probar tu micrófono", + "micGain": "Volumen del micrófono", + "systemGain": "Volumen del ordenador", "noMicrophones": "No se encontró ningún micrófono", "preview": "Vista previa", "previewUnavailable": "Vista previa no disponible", diff --git a/src/i18n/locales/fr/launch.json b/src/i18n/locales/fr/launch.json index b768831edf..4337bc039c 100644 --- a/src/i18n/locales/fr/launch.json +++ b/src/i18n/locales/fr/launch.json @@ -99,6 +99,8 @@ "close": "Fermer", "micLevel": "Niveau d'entrée", "micHint": "Parlez pour tester votre microphone", + "micGain": "Volume du micro", + "systemGain": "Volume de l'ordinateur", "noMicrophones": "Aucun microphone détecté", "preview": "Aperçu", "previewUnavailable": "Aperçu indisponible", diff --git a/src/i18n/locales/it/launch.json b/src/i18n/locales/it/launch.json index cad6d15516..bcb3519c57 100644 --- a/src/i18n/locales/it/launch.json +++ b/src/i18n/locales/it/launch.json @@ -99,6 +99,8 @@ "close": "Chiudi", "micLevel": "Livello di ingresso", "micHint": "Parla per testare il microfono", + "micGain": "Volume del microfono", + "systemGain": "Volume del computer", "noMicrophones": "Nessun microfono trovato", "preview": "Anteprima", "previewUnavailable": "Anteprima non disponibile", diff --git a/src/i18n/locales/ja-JP/launch.json b/src/i18n/locales/ja-JP/launch.json index aa6d4b5677..f0487f182b 100644 --- a/src/i18n/locales/ja-JP/launch.json +++ b/src/i18n/locales/ja-JP/launch.json @@ -99,6 +99,8 @@ "close": "閉じる", "micLevel": "入力レベル", "micHint": "マイクをテストするには話してください", + "micGain": "マイク音量", + "systemGain": "パソコンの音量", "noMicrophones": "マイクが見つかりません", "preview": "プレビュー", "previewUnavailable": "プレビューを利用できません", diff --git a/src/i18n/locales/ko-KR/launch.json b/src/i18n/locales/ko-KR/launch.json index 455a84146a..43c1e96633 100644 --- a/src/i18n/locales/ko-KR/launch.json +++ b/src/i18n/locales/ko-KR/launch.json @@ -99,6 +99,8 @@ "close": "닫기", "micLevel": "입력 레벨", "micHint": "마이크를 테스트하려면 말해 보세요", + "micGain": "마이크 볼륨", + "systemGain": "컴퓨터 소리 볼륨", "noMicrophones": "마이크를 찾을 수 없습니다", "preview": "미리 보기", "previewUnavailable": "미리 보기를 사용할 수 없습니다", diff --git a/src/i18n/locales/nl/launch.json b/src/i18n/locales/nl/launch.json index 0b668ab14b..771191b7e0 100644 --- a/src/i18n/locales/nl/launch.json +++ b/src/i18n/locales/nl/launch.json @@ -99,6 +99,8 @@ "close": "Sluiten", "micLevel": "Ingangsniveau", "micHint": "Spreek om je microfoon te testen", + "micGain": "Microfoonvolume", + "systemGain": "Computervolume", "noMicrophones": "Geen microfoon gevonden", "preview": "Voorbeeld", "previewUnavailable": "Voorbeeld niet beschikbaar", diff --git a/src/i18n/locales/pl/launch.json b/src/i18n/locales/pl/launch.json index 3ebecc4296..df4761a85e 100644 --- a/src/i18n/locales/pl/launch.json +++ b/src/i18n/locales/pl/launch.json @@ -99,6 +99,8 @@ "close": "Zamknij", "micLevel": "Poziom wejścia", "micHint": "Mów, aby sprawdzić mikrofon", + "micGain": "Głośność mikrofonu", + "systemGain": "Głośność dźwięku komputera", "noMicrophones": "Nie znaleziono mikrofonu", "preview": "Podgląd", "previewUnavailable": "Podgląd niedostępny", diff --git a/src/i18n/locales/pt-BR/launch.json b/src/i18n/locales/pt-BR/launch.json index 893576b96f..0403997408 100644 --- a/src/i18n/locales/pt-BR/launch.json +++ b/src/i18n/locales/pt-BR/launch.json @@ -99,6 +99,8 @@ "close": "Fechar", "micLevel": "Nível de entrada", "micHint": "Fale para testar seu microfone", + "micGain": "Volume do microfone", + "systemGain": "Volume do computador", "noMicrophones": "Nenhum microfone encontrado", "preview": "Pré-visualização", "previewUnavailable": "Pré-visualização indisponível", diff --git a/src/i18n/locales/ru/launch.json b/src/i18n/locales/ru/launch.json index a34023430a..912322a687 100644 --- a/src/i18n/locales/ru/launch.json +++ b/src/i18n/locales/ru/launch.json @@ -99,6 +99,8 @@ "close": "Закрыть", "micLevel": "Уровень входа", "micHint": "Скажите что-нибудь, чтобы проверить микрофон", + "micGain": "Громкость микрофона", + "systemGain": "Громкость звука компьютера", "noMicrophones": "Микрофон не найден", "preview": "Предпросмотр", "previewUnavailable": "Предпросмотр недоступен", diff --git a/src/i18n/locales/tr/launch.json b/src/i18n/locales/tr/launch.json index 929709e362..99ebf69a56 100644 --- a/src/i18n/locales/tr/launch.json +++ b/src/i18n/locales/tr/launch.json @@ -99,6 +99,8 @@ "close": "Kapat", "micLevel": "Giriş seviyesi", "micHint": "Mikrofonunuzu test etmek için konuşun", + "micGain": "Mikrofon ses düzeyi", + "systemGain": "Bilgisayar ses düzeyi", "noMicrophones": "Mikrofon bulunamadı", "preview": "Önizleme", "previewUnavailable": "Önizleme kullanılamıyor", diff --git a/src/i18n/locales/vi/launch.json b/src/i18n/locales/vi/launch.json index fe3ce29fa3..723c7f1e29 100644 --- a/src/i18n/locales/vi/launch.json +++ b/src/i18n/locales/vi/launch.json @@ -99,6 +99,8 @@ "close": "Đóng", "micLevel": "Mức đầu vào", "micHint": "Hãy nói để kiểm tra micrô", + "micGain": "Âm lượng micrô", + "systemGain": "Âm lượng máy tính", "noMicrophones": "Không tìm thấy micrô", "preview": "Xem trước", "previewUnavailable": "Không thể xem trước", diff --git a/src/i18n/locales/zh-CN/launch.json b/src/i18n/locales/zh-CN/launch.json index b53025f9c4..168146dc3f 100644 --- a/src/i18n/locales/zh-CN/launch.json +++ b/src/i18n/locales/zh-CN/launch.json @@ -99,6 +99,8 @@ "close": "关闭", "micLevel": "输入音量", "micHint": "说话试试,检查麦克风是否正常", + "micGain": "麦克风音量", + "systemGain": "电脑声音音量", "noMicrophones": "未找到麦克风", "preview": "预览", "previewUnavailable": "预览不可用", diff --git a/src/i18n/locales/zh-TW/launch.json b/src/i18n/locales/zh-TW/launch.json index 03d134c201..a7fc436d7f 100644 --- a/src/i18n/locales/zh-TW/launch.json +++ b/src/i18n/locales/zh-TW/launch.json @@ -99,6 +99,8 @@ "close": "關閉", "micLevel": "輸入音量", "micHint": "說話以測試麥克風", + "micGain": "麥克風音量", + "systemGain": "電腦聲音音量", "noMicrophones": "找不到麥克風", "preview": "預覽", "previewUnavailable": "預覽無法使用", diff --git a/src/lib/audioMix.test.ts b/src/lib/audioMix.test.ts index ba2443f001..6b7d397b38 100644 --- a/src/lib/audioMix.test.ts +++ b/src/lib/audioMix.test.ts @@ -125,6 +125,50 @@ describe("mixAudioTracks", () => { expect(result.track).toBeNull(); }); + it("applies the user's mic trim on top of the automatic mix gain", () => { + const fakeCtx = stubAudioContext(0.5); + + const result = mixAudioTracks({ + systemAudioTrack: track("sys"), + micAudioTrack: track("mic"), + micGainDb: 6, + }); + + expect(result.context).toBe(fakeCtx); + const [gain] = fakeCtx.createGain.mock.results.map((r) => r.value) as FakeGainNode[]; + // +6 dB is ~1.9953; the boost still applies because system audio is present. + expect(gain.gain.linearRampToValueAtTime).toHaveBeenCalledWith( + MIC_GAIN_BOOST * 10 ** (6 / 20), + 0.5 + MIC_FADE_IN_S, + ); + }); + + it("routes system-only audio through a GainNode when the user trims it", () => { + const fakeCtx = stubAudioContext(0.5); + + const result = mixAudioTracks({ + systemAudioTrack: track("sys"), + micAudioTrack: null, + systemGainDb: -6, + }); + + expect(result.context).toBe(fakeCtx); + const [trim] = fakeCtx.createGain.mock.results.map((r) => r.value) as FakeGainNode[]; + expect(trim.gain.value).toBeCloseTo(10 ** (-6 / 20), 6); + expect(result.track).toEqual({ kind: "audio", id: "dest-track" }); + }); + + it("keeps system-only audio verbatim at unity trim", () => { + const system = track("sys"); + const result = mixAudioTracks({ + systemAudioTrack: system, + micAudioTrack: null, + systemGainDb: 0, + }); + expect(result.context).toBeNull(); + expect(result.track).toBe(system); + }); + it("creates an AudioContext, fades the mic gain from 0 to MIC_GAIN_BOOST over MIC_FADE_IN_S, and returns the destination track when both inputs are present", () => { const fakeCtx = stubAudioContext(0.5); diff --git a/src/lib/audioMix.ts b/src/lib/audioMix.ts index d17980aacd..1935bd2cc7 100644 --- a/src/lib/audioMix.ts +++ b/src/lib/audioMix.ts @@ -24,9 +24,23 @@ export function nativeMicrophoneGain(systemAudioEnabled: boolean): number { */ export const MIC_FADE_IN_S = 0.02; +/** Bounds of the user's pre-recording volume trims, shared by the settings + * validation and the HUD sliders so both ends agree on what a legal value is. */ +export const RECORDING_GAIN_DB_MIN = -24; +export const RECORDING_GAIN_DB_MAX = 12; + +/** dB to the linear scalar the mixers multiply by (native helper and GainNode alike). */ +export function recordingGainScalar(gainDb: number): number { + return 10 ** (gainDb / 20); +} + export type MixAudioTracksInput = { systemAudioTrack?: MediaStreamTrack | null | undefined; micAudioTrack?: MediaStreamTrack | null | undefined; + /** User trim on the microphone, dB on top of the automatic mix gain. */ + micGainDb?: number; + /** User trim on the computer audio, dB. */ + systemGainDb?: number; }; export type MixAudioTracksResult = { @@ -46,25 +60,46 @@ export type MixAudioTracksResult = { * sit over system audio; on its own it rides at unity so mic-only loudness is unchanged. * Caller owns the returned `context` and must `.close()` it on teardown. * - * - System audio only: returned verbatim, no AudioContext created (no warm-up click). + * - System audio only at unity trim: returned verbatim, no AudioContext created + * (no warm-up click). A non-unity system trim routes it through a GainNode. * - Neither present: returns `{ context: null, track: null }`. */ export function mixAudioTracks({ systemAudioTrack, micAudioTrack, + micGainDb = 0, + systemGainDb = 0, }: MixAudioTracksInput): MixAudioTracksResult { + const systemGain = recordingGainScalar(systemGainDb); if (!micAudioTrack) { - return { context: null, track: systemAudioTrack ?? null }; + if (!systemAudioTrack || systemGain === 1) { + return { context: null, track: systemAudioTrack ?? null }; + } + const context = new AudioContext(); + const destination = context.createMediaStreamDestination(); + const systemSource = context.createMediaStreamSource(new MediaStream([systemAudioTrack])); + const trim = context.createGain(); + trim.gain.value = systemGain; + systemSource.connect(trim).connect(destination); + return { context, track: destination.stream.getAudioTracks()[0] }; } const context = new AudioContext(); const destination = context.createMediaStreamDestination(); if (systemAudioTrack) { const systemSource = context.createMediaStreamSource(new MediaStream([systemAudioTrack])); - systemSource.connect(destination); + if (systemGain === 1) { + systemSource.connect(destination); + } else { + const trim = context.createGain(); + trim.gain.value = systemGain; + systemSource.connect(trim).connect(destination); + } } - // Unity when the mic is on its own; boosted only when competing with system audio. - const micTargetGain = nativeMicrophoneGain(Boolean(systemAudioTrack)); + // Unity when the mic is on its own; boosted only when competing with system + // audio. The user's trim multiplies whichever base applies. + const micTargetGain = + nativeMicrophoneGain(Boolean(systemAudioTrack)) * recordingGainScalar(micGainDb); const micSource = context.createMediaStreamSource(new MediaStream([micAudioTrack])); const micGain = context.createGain(); micGain.gain.setValueAtTime(0, context.currentTime); diff --git a/src/lib/nativeLinuxRecording.ts b/src/lib/nativeLinuxRecording.ts index 2a1288864b..1ba69e4693 100644 --- a/src/lib/nativeLinuxRecording.ts +++ b/src/lib/nativeLinuxRecording.ts @@ -26,6 +26,8 @@ export type NativeLinuxRecordingRequest = { audio: { system: { enabled: boolean; + /** Linear user trim; absent means unity. */ + gain?: number; }; microphone: { enabled: boolean; diff --git a/src/lib/nativeMacRecording.ts b/src/lib/nativeMacRecording.ts index 257035638b..8ed796e4ed 100644 --- a/src/lib/nativeMacRecording.ts +++ b/src/lib/nativeMacRecording.ts @@ -27,6 +27,8 @@ export type NativeMacRecordingRequest = { audio: { system: { enabled: boolean; + /** Linear user trim; absent means unity. */ + gain?: number; }; microphone: { enabled: boolean; diff --git a/src/lib/nativeWindowsRecording.ts b/src/lib/nativeWindowsRecording.ts index 5d5d92b315..2d60cbf5e0 100644 --- a/src/lib/nativeWindowsRecording.ts +++ b/src/lib/nativeWindowsRecording.ts @@ -17,6 +17,8 @@ export type NativeWindowsRecordingRequest = { audio: { system: { enabled: boolean; + /** Linear user trim; absent means unity (older request builders). */ + gain?: number; }; microphone: { enabled: boolean; diff --git a/src/native/browserShim.ts b/src/native/browserShim.ts index 7643f32969..60a99e2cd3 100644 --- a/src/native/browserShim.ts +++ b/src/native/browserShim.ts @@ -57,6 +57,8 @@ type ShimRecordingPrefs = { camDeviceId: string | null; camDeviceName: string | null; systemAudioEnabled: boolean; + micGainDb: number; + systemAudioGainDb: number; cursorCaptureMode: "editable-overlay" | "system"; autoZoomEnabled: boolean; }; @@ -69,6 +71,8 @@ let shimRecordingPrefs: ShimRecordingPrefs = { camDeviceId: null, camDeviceName: null, systemAudioEnabled: false, + micGainDb: 0, + systemAudioGainDb: 0, cursorCaptureMode: "editable-overlay", autoZoomEnabled: true, }; From 2661dd81a4cda183b25a59b5eeefcd2bec82b731 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E4=B8=80=E4=B8=AA=E6=88=90=E7=86=9F=E7=9A=84=E5=89=AA?= =?UTF-8?q?=E8=BE=91=E7=8C=BF?= <297668904+akahoz94@users.noreply.github.com> Date: Fri, 20 Nov 2026 05:49:00 +0800 Subject: [PATCH 2/3] fix(audio): apply pre-record gain trims without distortion or stale values CodeRabbit review follow-ups for the volume-trims PR: - pipewire: keep an explicit gain of 0 (silence); only an absent gain falls back to unity, for both system and microphone sources. - wgc mixer: accumulate the gain-scaled system and microphone sources in a wider double domain and clamp the combined result once, so a boosted source cannot hit the rail before the opposite polarity is added. - HUD: lock gain controls from the record press through the countdown (including the Linux portal-preparation window) until recording starts; the countdown captures an earlier startRecording closure, so a trim changed mid-countdown would show but be ignored by the take. - linux IPC: include the system gain in the prepared-session key so a changed gain invalidates reuse of a helper spawned at the old gain. --- electron/ipc/handlers.ts | 1 + electron/native/pipewire-capture/src/main.rs | 5 +- .../wgc-capture/src/audio_sample_utils.cpp | 83 +++++++++++++++++-- .../src/audio_sample_utils_test.cpp | 22 +++++ src/components/launch/LaunchWindow.test.tsx | 10 +++ src/components/launch/LaunchWindow.tsx | 7 +- src/hooks/useScreenRecorder.ts | 16 +++- 7 files changed, 131 insertions(+), 13 deletions(-) diff --git a/electron/ipc/handlers.ts b/electron/ipc/handlers.ts index a6668a1d18..27298f1182 100644 --- a/electron/ipc/handlers.ts +++ b/electron/ipc/handlers.ts @@ -1010,6 +1010,7 @@ function captureSettingsOf(request: NativeLinuxRecordingRequest): string { fps: request.video?.fps ?? null, bitrate: request.video?.bitrate ?? null, system: request.audio?.system?.enabled ?? false, + systemGain: request.audio?.system?.gain ?? 1, microphone: request.audio?.microphone?.enabled ?? false, deviceName: request.audio?.microphone?.deviceName ?? null, gain: request.audio?.microphone?.gain ?? null, diff --git a/electron/native/pipewire-capture/src/main.rs b/electron/native/pipewire-capture/src/main.rs index 732139dc15..6d44407134 100644 --- a/electron/native/pipewire-capture/src/main.rs +++ b/electron/native/pipewire-capture/src/main.rs @@ -201,7 +201,8 @@ impl Request { // The default SINK's monitor: what is being played, not what a // microphone hears. capture_sink: true, - gain: audio.system.gain.filter(|gain| *gain > 0.0).unwrap_or(1.0), + // An explicit 0 means silence; only an absent gain falls back to unity. + gain: audio.system.gain.unwrap_or(1.0), bitrate: DEFAULT_AUDIO_BITRATE, }); } @@ -210,7 +211,7 @@ impl Request { label: "microphone", target: audio.microphone.device_name.clone(), capture_sink: false, - gain: audio.microphone.gain.filter(|gain| *gain > 0.0).unwrap_or(1.0), + gain: audio.microphone.gain.unwrap_or(1.0), bitrate: DEFAULT_AUDIO_BITRATE, }); } diff --git a/electron/native/wgc-capture/src/audio_sample_utils.cpp b/electron/native/wgc-capture/src/audio_sample_utils.cpp index 8964a0c5c2..8763f026b8 100644 --- a/electron/native/wgc-capture/src/audio_sample_utils.cpp +++ b/electron/native/wgc-capture/src/audio_sample_utils.cpp @@ -879,6 +879,68 @@ void mixAudioInPlace( } } +// Mixes up to two gain-scaled sources into a destination that starts as +// silence, accumulating in the destination's native numeric domain and clamping +// the COMBINED result exactly once. Applying each gain through a separate +// mixAudioInPlace clamps a boosted source to the rail before an +// opposite-polarity source can cancel it (clamp-then-sum); the wider double +// accumulator here preserves that headroom (sum-then-clamp). +// The destination is zero-filled by the caller, so it contributes nothing. +void mixTwoSourcesIntoSilence( + std::vector& destination, + const BYTE* systemSource, + double systemGain, + const BYTE* microphoneSource, + double microphoneGain, + const AudioInputFormat& format) { + const size_t bytesPer = bytesPerSample(format); + if (destination.empty() || bytesPer == 0) { + return; + } + const auto readRaw = [&](const BYTE* src, size_t index) -> double { + if (!src) { + return 0.0; + } + if (isFloatFormat(format)) { + return static_cast(reinterpret_cast(src)[index]); + } + if (isPcmFormat(format, 16)) { + return static_cast(reinterpret_cast(src)[index]); + } + if (isPcmFormat(format, 32)) { + return static_cast(reinterpret_cast(src)[index]); + } + return 0.0; + }; + const size_t sampleCount = destination.size() / bytesPer; + if (isFloatFormat(format)) { + auto* output = reinterpret_cast(destination.data()); + for (size_t i = 0; i < sampleCount; ++i) { + const double sum = + readRaw(systemSource, i) * systemGain + readRaw(microphoneSource, i) * microphoneGain; + output[i] = static_cast(std::clamp(sum, -1.0, 1.0)); + } + return; + } + if (isPcmFormat(format, 16)) { + auto* output = reinterpret_cast(destination.data()); + for (size_t i = 0; i < sampleCount; ++i) { + const double sum = + readRaw(systemSource, i) * systemGain + readRaw(microphoneSource, i) * microphoneGain; + output[i] = clampTo(sum); + } + return; + } + if (isPcmFormat(format, 32)) { + auto* output = reinterpret_cast(destination.data()); + for (size_t i = 0; i < sampleCount; ++i) { + const double sum = + readRaw(systemSource, i) * systemGain + readRaw(microphoneSource, i) * microphoneGain; + output[i] = clampTo(sum); + } + } +} + namespace { // How far behind real time AudioMixer writes; see mixLoop. Wider than the capture @@ -1123,7 +1185,8 @@ void AudioMixer::mixLoop() { const size_t chunkBytes = static_cast(chunkFrames) * format_.blockAlign; const uint64_t cushionFrames = static_cast(format_.sampleRate) * MixerCushionMs / 1000; std::vector mixedChunk; - std::vector sourceChunk; + std::vector systemChunk; + std::vector microphoneChunk; // This loop's copy of `clockStart_`, taken under the lock at the top of each // pass: beginTimeline and a resume move it from other threads. std::chrono::steady_clock::time_point clockStart; @@ -1147,17 +1210,19 @@ void AudioMixer::mixLoop() { std::scoped_lock lock(mutex_); mixedChunk.assign(chunkBytes, 0); if (includeSystem_) { - pop(systemQueue_, systemStarved_, sourceChunk, chunkBytes); - mixAudioInPlace( - mixedChunk, sourceChunk.data(), static_cast(sourceChunk.size()), format_, - systemGain_); + pop(systemQueue_, systemStarved_, systemChunk, chunkBytes); } if (includeMicrophone_) { - pop(microphoneQueue_, microphoneStarved_, sourceChunk, chunkBytes); - mixAudioInPlace( - mixedChunk, sourceChunk.data(), static_cast(sourceChunk.size()), format_, - microphoneGain_); + pop(microphoneQueue_, microphoneStarved_, microphoneChunk, chunkBytes); } + // Accumulate both sources in a wider double domain and clamp the + // sum once, so a boosted stream cannot hit the rail before the + // other stream's opposite polarity is added. + mixTwoSourcesIntoSilence( + mixedChunk, + includeSystem_ ? systemChunk.data() : nullptr, systemGain_, + includeMicrophone_ ? microphoneChunk.data() : nullptr, microphoneGain_, + format_); mixedFrames_ = emittedFrames_ + chunkFrames; } diff --git a/electron/native/wgc-capture/src/audio_sample_utils_test.cpp b/electron/native/wgc-capture/src/audio_sample_utils_test.cpp index 1d98eb1559..4e021be477 100644 --- a/electron/native/wgc-capture/src/audio_sample_utils_test.cpp +++ b/electron/native/wgc-capture/src/audio_sample_utils_test.cpp @@ -2850,6 +2850,28 @@ int main(int argc, char** argv) { std::cout << "GAIN_RAW unity " << detail << std::endl; expect("mixer-unity-gain-passthrough", std::abs(got - 29490) <= 2, detail); } + // (4b) A boosted system source that alone exceeds full scale must not be + // pinned to the rail before the opposite-polarity microphone is added: + // 0.8 * 2.0 - 0.5 * 1.0 = 1.1 -> clamped once at the rail. With + // 0.8 * 2.0 - 0.5 * 1.4 = 0.9 -> 29490, sum-then-clamp keeps the headroom + // that clamp-then-sum (which read 1.0 - 0.7 = 0.3) destroyed. + { + const int got = steadySample(true, 0.8, -0.5, 1.4, 2.0); + char detail[96]{}; + sprintf_s(detail, "got=%d want=29490", got); + std::cout << "GAIN_RAW boosted-system " << detail << std::endl; + expect("mixer-boosted-system-sum-before-clamp", std::abs(got - 29490) <= 2, detail); + } + // (4c) The same boost summed at the rail: 0.8 * 2.0 + 0.9 * 1.0 = 2.5 + // -> clamped once to 1.0 (32767). + { + const int got = steadySample(true, 0.8, 0.9, 1.0, 2.0); + char detail[96]{}; + sprintf_s(detail, "got=%d want=32767", got); + std::cout << "GAIN_RAW boosted-system-rail " << detail << std::endl; + expect("mixer-boosted-system-clamps-once-at-rail", got >= 32766, detail); + } + // (4) The system stream carries its own gain: 0.8 * 0.5 + 0.3 = 0.7 // -> 22937. A mixer that ignored systemGain would read 0.8 + 0.3 = 1.1 // clamped at the rail instead. diff --git a/src/components/launch/LaunchWindow.test.tsx b/src/components/launch/LaunchWindow.test.tsx index e72c7ac5a9..99c6007e3f 100644 --- a/src/components/launch/LaunchWindow.test.tsx +++ b/src/components/launch/LaunchWindow.test.tsx @@ -55,6 +55,7 @@ const recorderState = vi.hoisted(() => ({ paused: false, saving: false, elapsedSeconds: 0, + startingCountdown: false, toggleRecording: vi.fn(), togglePaused: vi.fn(), canPauseRecording: false, @@ -74,6 +75,10 @@ const recorderState = vi.hoisted(() => ({ setWebcamQuality: vi.fn(), systemAudioEnabled: false, setSystemAudioEnabled: vi.fn(), + micGainDb: 0, + setMicGainDb: vi.fn(), + systemAudioGainDb: 0, + setSystemAudioGainDb: vi.fn(), cursorCaptureMode: "editable-overlay", setCursorCaptureMode: vi.fn(), presenterOverlay: null as { active: boolean; cameraInterrupted: boolean } | null, @@ -366,6 +371,11 @@ function resetLaunchMocks() { recorderState.value.dismissSoftwareEncoderFallbackNotice.mockClear(); recorderState.value.presenterOverlay = null; recorderState.value.recording = false; + recorderState.value.startingCountdown = false; + recorderState.value.micGainDb = 0; + recorderState.value.setMicGainDb.mockClear(); + recorderState.value.systemAudioGainDb = 0; + recorderState.value.setSystemAudioGainDb.mockClear(); recorderState.value.canPauseRecording = false; recorderState.value.microphoneEnabled = false; recorderState.value.setMicrophoneEnabled.mockClear(); diff --git a/src/components/launch/LaunchWindow.tsx b/src/components/launch/LaunchWindow.tsx index 02eb11c8f8..708238e45e 100644 --- a/src/components/launch/LaunchWindow.tsx +++ b/src/components/launch/LaunchWindow.tsx @@ -126,6 +126,7 @@ export function LaunchWindow() { paused, saving, elapsedSeconds, + startingCountdown, toggleRecording, togglePaused, canPauseRecording, @@ -199,7 +200,11 @@ export function LaunchWindow() { const isVertical = trayLayout === "vertical"; const isPopoverOpen = isLanguageMenuOpen || isDeviceSettingsOpen; - const controlsLocked = recording || saving; + // Lock from the record press through the countdown (including the Linux portal + // preparation window) until recording starts. The countdown captured an earlier + // `startRecording` closure, so a trim changed mid-countdown would show on the + // slider but be ignored by the take. + const controlsLocked = recording || saving || startingCountdown; const settingsTriggerRef = useRef(null); const languageTriggerRef = useRef(null); diff --git a/src/hooks/useScreenRecorder.ts b/src/hooks/useScreenRecorder.ts index 19e0630407..37cb6e5a9f 100644 --- a/src/hooks/useScreenRecorder.ts +++ b/src/hooks/useScreenRecorder.ts @@ -94,6 +94,8 @@ type UseScreenRecorderReturn = { paused: boolean; saving: boolean; elapsedSeconds: number; + /** True from the record press until the countdown takes over (portal prep window). */ + startingCountdown: boolean; toggleRecording: () => void; /** Starts recording with no countdown overlay. Used by the headless CLI runner. */ startRecordingImmediately: () => Promise; @@ -452,6 +454,7 @@ export function useScreenRecorder(): UseScreenRecorderReturn { const countdownStarting = useRef(null); const cursorAccessibilityWarningShown = useRef(false); const [countdownActive, setCountdownActive] = useState(false); + const [startingCountdown, setStartingCountdown] = useState(false); const webcamReady = useRef(false); const webcamAcquireId = useRef(0); const canPauseRecording = @@ -1784,9 +1787,17 @@ export function useScreenRecorder(): UseScreenRecorderReturn { } const starting = {}; countdownStarting.current = starting; + // Publish a render-visible lock from this press onward, covering the Linux + // portal-preparation window before `countdownActive` is set. A gain edit + // there would move the slider but be missed by the `startRecording` + // closure this countdown captured. Kept separate from `countdownActive`: + // that flag gates press-to-cancel and must stay false until the countdown + // is actually showing (see #1340, the double-picker regression). + setStartingCountdown(true); try { await runRecordCountdown(); } finally { + setStartingCountdown(false); if (countdownStarting.current === starting) { countdownStarting.current = null; } @@ -1907,7 +1918,9 @@ export function useScreenRecorder(): UseScreenRecorderReturn { return; } - // From here `countdownActive` takes over, and a press cancels the countdown. + // From here `countdownActive` takes over: a press cancels the countdown. + // `startingCountdown` stays true until the whole countdown unwinds, so the + // HUD lock has no gap between the portal wait and the visible countdown. countdownStarting.current = null; setCountdownActive(true); @@ -2698,6 +2711,7 @@ export function useScreenRecorder(): UseScreenRecorderReturn { paused, saving, elapsedSeconds, + startingCountdown, toggleRecording, startRecordingImmediately: () => startRecording(), togglePaused, From c1e65ba33e9119356e6e86b0da2c9214491a27e8 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E4=B8=80=E4=B8=AA=E6=88=90=E7=86=9F=E7=9A=84=E5=89=AA?= =?UTF-8?q?=E8=BE=91=E7=8C=BF?= <297668904+akahoz94@users.noreply.github.com> Date: Fri, 20 Nov 2026 19:54:07 +0800 Subject: [PATCH 3/3] fix(audio): scope the countdown gain lock to its owning run CodeRabbit follow-up: a cancelled countdown's finally released the HUD gain lock unconditionally, so a countdown restarted within the same tick could unlock its controls while the new take still used the old startRecording closure. Gate release on an ownership token so only the current run can drop the lock. Adds a regression that cancels and restarts before the stale finally settles. --- .../useScreenRecorder.startCancel.test.tsx | 46 +++++++++++++++++++ src/hooks/useScreenRecorder.ts | 19 +++++++- 2 files changed, 64 insertions(+), 1 deletion(-) diff --git a/src/hooks/useScreenRecorder.startCancel.test.tsx b/src/hooks/useScreenRecorder.startCancel.test.tsx index e5d714bbcc..bc6fbf6ded 100644 --- a/src/hooks/useScreenRecorder.startCancel.test.tsx +++ b/src/hooks/useScreenRecorder.startCancel.test.tsx @@ -208,3 +208,49 @@ describe("a failed native start never leaves the webcam recording", () => { expect(handle.discard).not.toHaveBeenCalled(); }); }); + +describe("countdown gain lock belongs to the latest run", () => { + it("does not unlock when a cancelled countdown's finally settles after a new one starts", async () => { + stubElectronAPI("darwin"); + + // Hold run A at its overlay show; run B's show resolves immediately. + let releaseFirstShow!: () => void; + const firstShow = new Promise((resolve) => { + releaseFirstShow = () => resolve(true); + }); + let showCalls = 0; + api.showCountdownOverlay.mockImplementation(() => { + showCalls += 1; + return showCalls === 1 ? firstShow : Promise.resolve(true); + }); + + const view = renderHook(() => useScreenRecorder()); + await settle(); + + // A presses record, reaches the visible countdown, then hangs on its show. + await act(async () => { + view.result.current.toggleRecording(); + }); + await settle(); + expect(view.result.current.startingCountdown).toBe(true); + expect(showCalls).toBe(1); + + // Cancel A, then immediately press again to start B while A's show is pending. + await act(async () => { + view.result.current.toggleRecording(); + }); + await settle(); + await act(async () => { + view.result.current.toggleRecording(); + }); + await settle(); + expect(showCalls).toBe(2); + + // A's stale show now settles; its finally must not drop B's lock even though + // A's own countdown is long dead. The lock belongs to the newest run. + await act(async () => { + releaseFirstShow(); + }); + expect(view.result.current.startingCountdown).toBe(true); + }); +}); diff --git a/src/hooks/useScreenRecorder.ts b/src/hooks/useScreenRecorder.ts index 37cb6e5a9f..48ffabbc8c 100644 --- a/src/hooks/useScreenRecorder.ts +++ b/src/hooks/useScreenRecorder.ts @@ -452,6 +452,14 @@ export function useScreenRecorder(): UseScreenRecorderReturn { * grant the user had just given the first (#1340). */ const countdownStarting = useRef(null); + /** + * Ownership token for the HUD gain lock. `countdownStarting` is cleared as soon + * as the countdown shows (see above), but the lock must survive through the + * visible countdown until the whole start unwinds. A stale run's `finally` must + * not drop a newer run's lock when a countdown is cancelled and restarted, so + * release is gated on this token rather than `countdownStarting`. + */ + const startingCountdownOwner = useRef(null); const cursorAccessibilityWarningShown = useRef(false); const [countdownActive, setCountdownActive] = useState(false); const [startingCountdown, setStartingCountdown] = useState(false); @@ -1787,6 +1795,8 @@ export function useScreenRecorder(): UseScreenRecorderReturn { } const starting = {}; countdownStarting.current = starting; + const owner = {}; + startingCountdownOwner.current = owner; // Publish a render-visible lock from this press onward, covering the Linux // portal-preparation window before `countdownActive` is set. A gain edit // there would move the slider but be missed by the `startRecording` @@ -1797,7 +1807,14 @@ export function useScreenRecorder(): UseScreenRecorderReturn { try { await runRecordCountdown(); } finally { - setStartingCountdown(false); + // Only the run that still owns the lock may release it. If countdown A + // was cancelled and B started before A's `await` settled, A's finally + // must leave B's lock intact; otherwise a gain edit during B would move + // the slider but be ignored by B's captured startRecording closure. + if (startingCountdownOwner.current === owner) { + startingCountdownOwner.current = null; + setStartingCountdown(false); + } if (countdownStarting.current === starting) { countdownStarting.current = null; }