flo-bit.dev / youtube-speaker-separation
[READ-ONLY] Mirror of https://github.com/flo-bit/youtube-speaker-separation. simple python script that outputs separate audio files for each speaker in a youtube video, using whisper on replicate
speaker-diarization speech-to-text text-to-speech voice-cloning whisper youtube
0
Fork 0

Configure Feed

Select the types of activity you want to include in your feed.

Python 100.0%
2 1 0

Clone this repository

https://tangled.org/flo-bit.dev/youtube-speaker-separation https://tangled.org/did:plc:4youxojjuvwv3sngqsb5zqg3
git@tangled.org:flo-bit.dev/youtube-speaker-separation git@tangled.org:did:plc:4youxojjuvwv3sngqsb5zqg3

For self-hosted knots, clone URLs may differ based on your setup.


Download tar.gz Download .zip
Readme.md

youtube speaker separation#

this script uses whisper on replicate to download the audio and then separate the speakers from a youtube video.

  • input: youtube url

  • output: audio files for each speaker, with all spoken parts of that speaker cut together.

this output can then be used to create ai voices for each speaker, for example with instant voice cloning by elevenlabs (note: elevenlabs only official allows this for voices that you own the rights to, i don't take any responsibility for any illegal use of this software.)

how to use#

create a virtual environment and install the requirements:

python3 -m venv .myenv
source .myenv/bin/activate
pip3 install -r requirements.txt

copy and rename .env.sample to .env and fill the REPLICATE_API_TOKEN field with your replicate api token.

run the script with a youtube url as argument:

python3 script.py "youtube_url"

license#

mit