Version 3 - continuation shard of the video-to-AI-dataset tokenizer project.
Version 2 - continuation shard of the video-to-AI-dataset tokenizer project.
Production-ready pipeline (Python package video_vec2wav2_tokenizer, CLI command
video2dataset) that turns a folder of videos into clean AI training datasets
for speech recognition (ASR) and text-to-speech (TTS).
videos ──► audio (16 kHz mono PCM) ──► whisper transcript ──► clips ──►… See the full description on the dataset page:
https://huggingface.co/datasets/k9cli/video-vec2wav2-tokenizer-3.
