Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Experimental code: sound file preprocessing to optimize Whisper transcriptions without hallucinated texts
| Date | Stars |
|---|---|
| 2026-07-24 | 350 |
| 2026-07-25 | 350 |
| 2026-07-28 | 351 |
| 2026-07-30 | 351 |
| 2026-08-06 | 351 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# WhisperHallu Experimental code: sound file preprocessing to optimize Whisper transcriptions without hallucinated texts See this discussion: https://github.com/openai/whisper/discussions/679 # Main algo - remove noise by voice extraction using [Facebook Demucs](https://github.com/facebookresearch/demucs) or [Deezer Spleeter](https://github.com/deezer/spleeter). - remove silences, and normalize loudness with ffmpeg. - remove noise parts using [Silero VAD](https://github.com/snakers4/silero-vad). - add voice markers. - apply speech compressor (requires `ffmpeg` 4.4, while Google Colab is 4.2, it has to be upgraded, see below). - try to transcribe. If markers are present in output, transcription is OK. - if not, try to invert markers. If markers are present in output, transcription is OK. - if not, try without markers. # Processing options and parameters - use Whisper V1, V2 or V3 (V2 by default, because V3 seems bad with music). - beam_size (2 by default), patience, temperature. - process only a subpart of the input file (needs a post-processing of timestamp values). - various time stretching methods tested (see in-code comments. Needs a post-processing of timestamp values. It was an interesting suggested idea, but no real gain obtained on my side). - vocals remix (with or without speech normalization). - multiple final transcriptions (get multiple results, knowing Whisper is not stable from one run to an other, without doing pre-processing several times) # Complement May be used to produce "accurate transcriptions" for WhisperTimeSync:<br/> https://github.com/EtienneAb3d/WhisperTimeSync May be tested using NeuroSpell Dictaphone:<br/> https://neurospell.com/ WhisperHallu and WhisperTimeSync are used to extract vocals and lyrics in karaok-AI:<br/> https://github.com/EtienneAb3d/karaok-AI ChatMate is a complete versatile ChatGPT automation tool, including explanations to produce a SRT file translator to Chinese (as an example):<br/> https://github.com/EtienneAb3d/ChatMate # Google Colab Standard Whisper:<br/> https://colab.research.google.com/drive/1-GpXaNaGFXKX9VXl60JGVVrGO41t09KA?usp=sharing Faster Whisper:<br/> https://colab.research.google.com/drive/1RkvOtUTbUD5NVsRI4aKEqJO8BRo8BFIY?usp=sharing # Install **Check ffmpeg version >=4.4** ```sh ffmpeg -version Output should be: ================= ffmpeg version 4.4.3-0ubuntu1~20.04.sav2 Copyright (c) 2000-2022 the FFmpeg developers [...] Install latest: =============== sudo add-apt-repository -y ppa:savoury1/ffmpeg4 sudo apt-get -qq install -y ffmpeg ``` **Demucs (if used)** ```sh pip install -U demucs ``` **Spleeter (if used)** ```sh pip install spleeter ``` **Standard Whisper (if used)** ```sh sudo apt update && sudo apt install ffmpeg sudo apt install python3 sudo apt install python3-pip sudo apt install virtualenv virtualenv -p python3 ../venvWhisper . ../venvWhisper/bin/activate pip install -U openai-whisper pip3 install torchaudio ``` **Faster Whisper (if used in place of Whisper)** ```sh sudo apt update && sudo apt install ffmpeg sudo apt install python3 sudo apt install python3-pip sudo apt install virtualenv virtualenv -p python3 ../venvFasterWhisper . ../venvFasterWhisper/bin/activate git clone https://github.com/guillaumekln/faster-whisper.git cd faster-whisper/ pip install -e .[conversion] pip install -e . cd .. ct2-transformers-converter --model openai/whisper-medium --output_dir whisper-medium-ct2 --quantization float16 ct2-transformers-converter --model openai/whisper-large --output_dir whisper-large-ct2 --quantization float16 pip3 install torchaudio ``` **SM4T (if used in place of Whisper)** ```sh sudo apt update && sudo apt install ffmpeg sudo apt install python3 sudo apt install python3-pip sudo apt install virtualenv virtualenv -p python3 ../venvSM4T . ../venvSM4T/bin/activate git clone https://github.com/facebookresearch/seamless_communication.git cd seamless_communication/ pip install --upgrade pip pip install . m4t_
Excerpt of 5,170 characters
Read on GitHubWould you bet a product on this? Bounded 0–100 and slow moving.
matched fp:e08a542b00381a94, topic:text-to-speech, topic:asr, topic:whisper