# What is the best Speech to Text Local Model? — retained record

> Retained reddit record from reddit:LocalLLaMA.

Canonical: https://highsignal.app/data/records/1ff4e2260c7db927

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:LocalLLaMA)  published 2026-11-01 14:27 ingested 2026-11-02 02:41

[original source ↗](https://www.reddit.com/r/LocalLLaMA/comments/1xl4f4h/what_is_the_best_speech_to_text_local_model/)

## retained text



```
I ran tests with Fable on AI generated Audio. Nemotron 3.5 Transcribe came in first. but I was curious about anyone has used other models and compared against. this. 

My use case is Audio transcription of 3 unique voices in a semi noisy room (HVAC sounds)
```



document fetched 2026-11-02 00:36 · canonical https://reddit.com/r/LocalLLaMA/comments/1xl4f4h/what_is_the_best_speech_to_text_local_model
