Feature: Add SenseVoice model — non-autoregressive, 50+ languages, emotion detection
Maintainers usually reply within 2 days
Nobody has claimed this yet.
Assessment
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Newbie friendliness
- 48/100
- Issue type
- Feature
- Clarity
- Mostly clear
- Activity status
- Quiet
- Tech stack
- swift
- Domain
- machine-learning
Research direction
Start by reading FluidVoice's existing Parakeet and other sherpa-onnx model integrations to understand how models are exposed as options and run on Apple Silicon. Then determine how the SenseVoice model and its emotion or audio-event outputs fit that flow. Done means SenseVoice is available as a model option and can transcribe audio; the issue does not specify whether emotion and audio-event detection are in scope.
Written by the indexing model from the issue text.
Description
Feature Request
FluidVoice's multi-model architecture is excellent — supporting Parakeet, Whisper, Cohere, and Apple Speech. I'd like to suggest adding SenseVoice as another model option.
Why SenseVoice?
| Feature | SenseVoice Small | Whisper Large | Parakeet v3 |
|---|---|---|---|
| Architecture | Non-autoregressive | Autoregressive | CTC |
| Parameters | 234M | 1.5B | 600M |
| Speed | ~170× realtime | ~15× realtime | ~50× realtime |
| Languages | 50+ | 99 | 25 |
| Extras | Emotion, audio events | — | — |
Key benefits for FluidVoice users:
- Non-autoregressive: Single forward pass — fastest possible inference, no beam search
- Compact: 234M params, ~400MB download — smaller than any Whisper variant
- Emotion + audio events: Detects laughter, applause, music alongside transcription
- Apple Silicon optimized: Available via sherpa-onnx with CoreML/Metal support — same integration path as your existing models
Integration Path
SenseVoice is already available through sherpa-onnx (which FluidVoice uses for other models). The ONNX model can run on Apple Silicon via CoreML:
Model: sherpa-onnx-sense-voice-zh-en-ja-ko-yue-2024-07-17
Size: ~400MB
Languages: Chinese, English, Japanese, Korean, Cantonese + 45 more
References
- SenseVoice repo: https://github.com/FunAudioLLM/SenseVoice (8.3K ⭐)
- sherpa-onnx: https://github.com/k2-fsa/sherpa-onnx (already used in FluidVoice ecosystem)
- FunASR toolkit: https://github.com/modelscope/FunASR (16.6K ⭐)
Happy to help with integration details!
- Dominant language
- Swift
- Stars
- 11.9k
- Forks
- 863
- Avg merge
- 1d 16h
- Merged PRs (30d)
- 26
Getting set up
This project ships no dev container, Dockerfile or contributing guide, so setting up is up to you: start from its README, and see our first-contribution guide for the general steps.
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
More from altic-dev/FluidVoice
-
enhancement
Difficulty 2/5 1-3 hours Newbie friendliness 65/100
altic-dev/FluidVoice#1117 ·
Maintainers usually reply within 2 days
-
bug
Difficulty 4/5 3-5 days Newbie friendliness 45/100
altic-dev/FluidVoice#1119 ·
Maintainers usually reply within 2 days
-
enhancement
Difficulty 4/5 3-5 days Newbie friendliness 52/100
altic-dev/FluidVoice#1118 ·
Maintainers usually reply within 2 days
-
enhancement
Difficulty 3/5 1-2 days Newbie friendliness 55/100
altic-dev/FluidVoice#1116 · 1 comment ·
Maintainers usually reply within 2 days
-
[BUG] v1.6.11: dictation refuses to paste into the Muse macOS app (focus check reads AXButton); works in Chrome and GeminiPossibly taken @jeevvansh claimed this today. Openbug
Difficulty 3/5 1-2 days Newbie friendliness 35/100
altic-dev/FluidVoice#1109 · 1 comment ·
Maintainers usually reply within 2 days
All issues in altic-dev/FluidVoice
Similar issues
-
area: sidebar bug S2: major
Difficulty 2/5 1-3 hours Newbie friendliness 72/100
manaflow-ai/cmux#20281 · 1 comment ·
Maintainers usually reply within 1 day
-
If a snippet replacement starts with a newline, it truncates the first parts of the replacement.Open
Difficulty 2/5 1-3 hours Newbie friendliness 75/100
-
Difficulty 2/5 1-3 hours Newbie friendliness 80/100
margelo/react-native-vision-camera#4216 ·
Maintainers usually reply within 4 days
-
documentation good first issue
Difficulty 2/5 1-3 hours Newbie friendliness 78/100
-
bug
Difficulty 2/5 1-3 hours Newbie friendliness 62/100
compiler-explorer/compiler-explorer#9272 ·
Maintainers usually reply within 1 day