Audio Services

Decoding various audio/video formats

The SoundFingerprinting library decodes audio samples (or video frames) from the underlying media file as the first step in the processing pipeline. It has to extract AudioSamples for audio fingerprints (or Frames for video fingerprints).

The decoding is done by instances of the following interfaces that you can specify in the UsingServices method overload during fingerprint creation or querying:

• IAudioService - used for audio fingerprinting.
  • SoundFingerprintingAudioService
  • FFmpegAudioService
  • NAudioService
  • BassAudioService
• IVideoService - used for video fingerprinting.
  • FFmpegAudioService
• IMediaService - used for both audio and video fingerprinting.
  • FFmpegAudioService

If neither of these implementations meets your requirements (i.e., none of the classes can decode file formats specific to your use case), you can implement your own class and use it in the processing pipeline.

What service is suitable for you?

1. SoundFingerprintingAudioService comes bundled with the core SoundFingerprinting library. It supports only WAV files at the input. This implementation is known to alias the audio signal during downsampling. You can use this implementation only in testing environments. It is supported in any OS and is MIT-licensed.

2. FFmpegAudioService comes bundled with SoundFingerprinting.Emy package. FFmpegAudioService is also the only currently supported service that can be used to read video files, providing the ability to fingerprint video frames and audio samples. Best suited for the production environment due to speed and accuracy. It provides ways to read audio from any format that is supported by FFmpeg (that is almost any known format). It is also able to extract audio tracks directly from video files. To use FFmpegAudioService you need to have it installed on the machine that fingerprints the files:

   • Windows - download FFmpeg libraries for x64 platform. Place the files alongside your executable in the following folder: FFmpeg\bin\x64. The recommended version is ffmpeg-6.0-full_build-shared.
   • Debian 12 - apt-get install ffmpeg (only supported with SoundFingerprinting.Emy v9.x)
   • Debian 11 - apt-get install ffmpeg (only supported with SoundFingerprinting.Emy v8.x)
   • Alpine - apk add ffmpeg (verify with ffmpeg --version which version is installed).
   • MacOS - brew install ffmpeg@6
     • Install ffmpeg@6 for SoundFingerprinting.Emy v10.x
     • Install ffmpeg@5 for SoundFingerprinting.Emy v9.x
     • Install ffmpeg@4 for SoundFingerprinting.Emy v8.x

   SoundFingerprinting.Emy is free for non-commercial use. Please contact sergiu@emysound.com for details. Please follow the below version matrix.

SoundFingerprinting	SoundFingerprinting.Emy	FFmpeg
8.x	8.x	4.x
9.x	9.x	5.x
10.x	10.x	6.x
N/A	N/A	7.x

3. Deprecated NAudioService suitable for the Windows environment only. Available via SoundFingerprinting.Audio.NAudio NuGet package. Provides support for .mp3 audio file processing. MIT licensed.

4. Deprecated BassAudioService available for Windows and Unix environment via SoundFingerprinting.Audio.Bass NuGet package. Supports multiple audio formats (.wav, .ogg, .mp3, .flac). Bass is free for non-comercial use. Installation details can be found here.