[GH-ISSUE #3583] [FEAT]: Support phi 4 for Speech to Text #2316

Closed
opened 2026-02-22 18:29:08 -05:00 by yindo · 2 comments
Owner

Originally created by @MyButtermilk on GitHub (Apr 2, 2025).
Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3583

What would you like to see?

Phi 4 Instruct from Microsoft is really accurate in speech recognition and it is free because it is open source. It is the best open source model available according to the open ASR benchmark:
https://huggingface.co/spaces/hf-audio/open_asr_leaderboard

https://github.com/marketplace/models/azureml/Phi-4-multimodal-instruct

https://huggingface.co/microsoft/Phi-4-multimodal-instruct

Supporting this great model would be a benefit because it would significantly improve the speech recognition abilities.

Originally created by @MyButtermilk on GitHub (Apr 2, 2025). Original GitHub issue: https://github.com/Mintplex-Labs/anything-llm/issues/3583 ### What would you like to see? Phi 4 Instruct from Microsoft is really accurate in speech recognition and it is free because it is open source. It is the best open source model available according to the open ASR benchmark: https://huggingface.co/spaces/hf-audio/open_asr_leaderboard https://github.com/marketplace/models/azureml/Phi-4-multimodal-instruct https://huggingface.co/microsoft/Phi-4-multimodal-instruct Supporting this great model would be a benefit because it would significantly improve the speech recognition abilities.
yindo added the enhancementfeature request labels 2026-02-22 18:29:08 -05:00
yindo closed this issue 2026-02-22 18:29:08 -05:00
Author
Owner

@timothycarambat commented on GitHub (Apr 3, 2025):

TTS is not per model - it is per provider! So this would come in the form of supporting providers who, in turn, support phi 4 for this task and we can then leverage those providers for audio transcription 👍

@timothycarambat commented on GitHub (Apr 3, 2025): TTS is not per model - it is per _provider_! So this would come in the form of supporting providers who, in turn, support phi 4 for this task and we can then leverage those providers for audio transcription 👍
Author
Owner

@MyButtermilk commented on GitHub (Apr 3, 2025):

Open router supports it:
https://openrouter.ai/microsoft/phi-4-multimodal-instruct

Would that work? One needs to add a short system prompt.

Transcribe this audio. Return only the transcript with correct spelling, grammar and punctuation.

@MyButtermilk commented on GitHub (Apr 3, 2025): Open router supports it: https://openrouter.ai/microsoft/phi-4-multimodal-instruct Would that work? One needs to add a short system prompt. Transcribe this audio. Return only the transcript with correct spelling, grammar and punctuation.
yindo changed title from [FEAT]: Support phi 4 for Speech to Text to [GH-ISSUE #3583] [FEAT]: Support phi 4 for Speech to Text 2026-06-05 14:45:51 -04:00
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Mintplex-Labs/anything-llm#2316