Tiny Audio - a mazesmazes Collection

mazesmazes 's Collections

Tiny Audio

updated Oct 9

A speech-to-text system you can train yourself: model, live demo, and the models it builds on.

mazesmazes/tiny-audio

Automatic Speech Recognition • 0.6B • Updated 22 days ago • 291 • 1

Note The published model: transcripts, word timestamps, speaker labels
Running on Zero

Agents

3

Tiny Audio Demo

🎤

3

Fast English ASR with word timestamps and speaker labels

Note Live demo
ibm-granite/granite-speech-5.0-470m-turboctc

Automatic Speech Recognition • 0.5B • Updated 7 days ago • 54.4k • 119

Note Frozen speech encoder
Qwen/Qwen3.5-2B

Image-Text-to-Text • 2B • Updated Mar 2 • 4.63M • • 484

Note Decoder (frozen, with LoRA)