Open-Source Attributions
TTSAudit is built with the help of the following open-source projects. We are grateful to their authors and contributors.
SpeechBrain
Apache 2.0Open-source speech toolkit used for speaker verification via the ECAPA-TDNN model trained on VoxCeleb.
https://github.com/speechbrain/speechbrainFaster Whisper
MITCTranslate2-based reimplementation of OpenAI Whisper, used for automatic speech recognition and pace analysis.
https://github.com/SYSTRAN/faster-whisperOpenAI Whisper
MITGeneral-purpose speech recognition model. Faster Whisper is derived from this work.
https://github.com/openai/whisperPyTorch
BSD-3-ClauseOpen-source machine learning framework used as the foundation for model inference.
https://github.com/pytorch/pytorchtorchaudio
BSD-2-ClauseAudio processing library for PyTorch, used for audio loading and waveform manipulation.
https://github.com/pytorch/audiolibrosa
ISCPython library for audio and music signal analysis, used for spectral feature extraction in quality analysis.
https://github.com/librosa/librosaNumPy
BSD-3-ClauseFundamental package for numerical computing in Python.
https://github.com/numpy/numpySciPy
BSD-3-ClauseScientific computing library used for signal processing and statistical analysis.
https://github.com/scipy/scipyFlask
BSD-3-ClauseLightweight WSGI web framework used for the API layer.
https://github.com/pallets/flaskAngular
MITWeb application framework used for the TTSAudit frontend and dashboard.
https://github.com/angular/angularSentry SDKs
MITError monitoring and performance tracking SDKs for both frontend and backend.
https://github.com/getsentry/sentry-javascript