Convert audio to srt:

• cd batch-folder-audio-to-srt
• Run: python3 audio-to-srt.py "/mnt/c/filepath/to/audio.m4a" to convert .m4a audio to srt file.
• .srt file will be outputted to ./SrtFiles in same location where you ran the script.

Batch convert all audio files in folder to srt:

• cd batch-folder-audio-to-srt
• Run: python3 run-for-folder.py
• Input folder hardcoded in script.
• .srt file will be outputted to ./SrtFiles in same location where you ran the script.

Upload single srt file to algolia database

• cd upload-srt-to-algolia
• Fill out .env file based off .env-template
• Run python3 test.py to test out connection.
• Run python3 upload-srt-to-algolia.py "/mnt/c/Users/marti/My Cool Title [htUDG4re1u4]_subtitles.srt" to convert .srt file to json and upload to algolia DB.

---

Benchmarks

• new-auth-to-srt-fast
  • run with python3 convert.py "/mnt/d/Jerma985 Streams Audio/Livestream： TF2⧸CS：GO： April 14, 2012 (Pre Recorded) - Jerma985.m4a"
  • "Livestream： TF2⧸CS：GO： April 14, 2012 (Pre Recorded) - Jerma985.m4a" = 1:18:43 long
  • Took 18:27 to generate srt file.
  • For every minute of input audio, the script processes it in approximately 14.07 seconds.
  • So the 01:58:14 audio file (60+58=118 minutes total) "TF2⧸Tribes： Ascend Livestream! (Pre Recorded) [3qWKO3LVCxc].m4a" should take 118minutes*14.07 = 1660.26 seconds, which is 27.671 minutes.
  • python3 convert.py "/mnt/d/Jerma985 Streams Audio/TF2⧸Tribes： Ascend Livestream! (Pre Recorded) [3qWKO3LVCxc].m4a"

How to setup/use virt env

Windows: venv_name\Scripts\activate Mac/Linux: source venv_name/bin/activate Leave: deactivate pip freeze > requirements.txt pip install -r requirements.txt

Test 1

• python aaiTest.py > output.txt

Test 2

• python whisperx_example.py

---

https://www.youtube.com/watch?v=SAIsk0i7KgU&t=474s&ab_channel=PromptEngineering

Convert youtube to mp3

yt-dlp -f bestaudio -x --audio-format mp3 --audio-quality 0 --add-metadata https://www.youtube.com/watch?v=AL2IkW4JWl4

Convert mp3 to txt tracnscript

https://github.com/openai/whisper pip install -U openai-whisper sudo apt update && sudo apt install ffmpeg whisper audio.mp3 --model medium python3 -m whisper jerma\ GROWLS\ at\ chat\ and\ has\ a\ meltdown\ \[AL2IkW4JWl4\].mp3 --model medium

Convert audio to transcript with speaker identification

python aaiTest.py > output.txt

WhisperX Diarisation

• https://github.com/m-bain/whisperX

1. Install Conda win10: https://docs.conda.io/projects/miniconda/en/latest/
2. Init conda.exe: https://conda.io/projects/conda/en/latest/user-guide/getting-started.html
3. Follow steps: https://github.com/m-bain/whisperX?tab=readme-ov-file#1-create-python310-environment
4. make sure ffmpeg works from command line ex: compute_type="int8"