Data Card.md

Audio Samples Single Note Acoustic Synthesised

Dataset Card for Audio Samples

Table of Contents

• Dataset Description
  • Dataset Summary
  • Supported Tasks
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Source Data
  • Annotations
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
  • Other Known Limitations
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information

Dataset Description

• Homepage: [Needs More Information]
• Repository: https://github.com/RikMaxSpeed/SampleGen
• Paper: [Needs More Information]
• Leaderboard: [Needs More Information]
• Point of Contact: RikMaxSpeed@gmail.com

Dataset Summary

A collection of approx 1000 audio samples of single notes from acoustic or synthesised instruments, containing only the notes C3 and C4.

This data-set is intended to enable the generation of new samples, which can then be performed by a synthesiser.

The file names, which typically describe the samples, are in English.

Supported Tasks and Leaderboards

Imperial personal College CapStone project.

Languages

English, colloquial, with some inside musical industry jokes, for example spelling the famous "Rhodes" piano as "Roadz".

Dataset Structure

Data Instances

All files are in the .wav format, the majority have a sample rate of 44.1 kHz and a few are at 48 kHz.

Data Fields

None other than the file name from which a sample type or category could potentially be inferred.

Data Splits

The data-split is up to the user. Typically train=80%, test=20%.

Dataset Creation

Curation Rationale

The data-set was extracted from multiple sources, including:

• Free Wave Samples: public domain wave samples https://freewavesamples.com
• Music Radar: https://www.musicradar.com/news/sampleradar-free-essential-synth-samples
• Apple Logic (licensed software)
• Arturia Pigments (licensed software)

Although single-note samples are typically not copyrightable, these are all samples for which I, personally, hold a license.

This data-set is therefore not in the public-domain.

Source Data

Initial Data Collection and Normalization

The only normalisation here was to restrict the files to those containing the note C3 or C4. This could be expanded in future.

Who are the source language producers?

File names are in English. No further text is available.

Annotations

Annotation process

None.

Who are the annotators?

Not Applicable

Personal and Sensitive Information

These audio files contain no personal information. However some come from licensed software distributions, therefore this dataset is for my personal use only.

Considerations for Using the Data

Social Impact of Dataset

None.

Discussion of Biases

These are single-note samples of musical instruments from all over the world. However there is a strong bias towards instruments used in Western music, ie: piano, strings, brass, organ etc. Note: this dataset does not include audio-loops or sound effects.

Other Known Limitations

[Needs More Information]

Additional Information

Dataset Curators

Richard Meyer

Licensing Information

Parts of the dataset are sourced from Apple & Arturia, therefore this dataset is not publicly available.

Citation Information

[Needs More Information]