artisan-tts

Published: Dec 05, 2023

One-day build. Source code.

Allows you to make janky TTS using a set of samples. This is hand-crafted TTS — artisanal!

It’s a script that expects a folder of audio files and text files and allows the user to select audio files using the text file contents (which are presumably transcriptions). Then, it uses OpenAI’s whisper to transcript the audio. The user can then select a segment from that transcription. Finally, it joins all selected audio segments into a single file.