LocalClaw/TTS/MOSS-Audio 4B Instruct
Local ASR model

MOSS-Audio 4B Instruct

Catalogue summary: Apache-licensed audio-language model for speech, sound and audio understanding. The 4B Instruct tier combines a MOSS audio encoder with a Qwen3-4B backbone, supports transcription, speech captioning, audio QA and time-aware reasoning, and ships downloadable BF16 safetensors with official local PyTorch, Gradio and SGLang paths.

Repository editorial metadata; verify comparative claims in the linked upstream material.

GPU recommendedspeech-to-text transcription2 languagesApache 2.0
Choose an app

Start with Recommended. No terminal commands are shown.

Compare speech models
Open model filesRecommended verified starting pointOpen in UnslothLaunches Unsloth Desktop on this modelOpen official projectApplication source and guided setupOpen on Hugging FaceFiles, licence and model cardCheck my machineOptional: verify the fit with your saved hardware

Desktop app links require the app to be installed. If nothing opens, LocalClaw will show app-download and model-file fallbacks.

Catalogue quality
9.2/10
Catalogue speed
7.7/10
Model size
5.22B BF16 params; 10.5 GB HF repo
Voices
N/A (audio-language understanding, ASR and captioning)

Can MOSS-Audio 4B Instruct run locally?

MOSS-Audio 4B Instruct can run locally for offline speech-to-text. Use the verified setup options on this page; no terminal command is required to choose the right path.

Apache 2.0 license. Still verify upstream usage notes before shipping.

Upstream source
multilingualcontrollablelong-form

Audio profile

Cat. quality
9.2
Cat. speed
7.7
Audio
8.7

Best fit

MOSS-Audio 4B Instruct is best for offline transcription, speech indexing and local voice pipelines.

Hardware: gpu

Model details

Type
Local ASR model
Family
moss
Latency
medium
Formats
pytorchsafetensorssglang
Languages
en, zh
Context
MOSS-Audio encoder plus Qwen3-4B backbone, 5.22B BF16 safetensors, Python 3.12 local runtime, ffmpeg and PyTorch CUDA setup; evaluated on audio understanding, speech captioning, ASR and timestamp ASR

Install locally

01
Check runtimeConfirm the backend supports pytorch, safetensors, sglang on your machine.
02
Open recommended setupUse the app and model links above. LocalClaw does not expose a terminal command.
03
Test locallyRun a short private audio prompt before moving into production workflows.

Good for

  • speech-to-text transcription
  • GPU recommended local workflows
  • multilingual, controllable, long-form

Watch before shipping

  • Validate pronunciation, latency and artifacts with your own voice samples.
  • Review the upstream license and acceptable-use notes.
  • Benchmark on your target CPU, Apple Silicon or GPU setup.

Related TTS and speech models

OpenMOSS / MOSI.AI MOSS-Transcribe-Diarize 0.9B Local ASR model · Q 9.4 · Speed 8.4 OpenMOSS / MOSI.AI MOSS-TTS Local TTS model · Q 9.2 · Speed 6.8 OpenMOSS / MOSI.AI MOSS-TTS-Nano Local TTS model · Q 8.5 · Speed 9.7 FireRed Team FireRedAudio Local ASR model · Q 9.5 · Speed 6.6 KRAFTON AI A.X K2 Raon-Speech 21B-A3B Local ASR model · Q 9.4 · Speed 6.4 NVIDIA Parakeet TDT 0.6B v3 Local ASR model · Q 9.5 · Speed 9.7 NVIDIA Nemotron 3.5 ASR Streaming 0.6B Local ASR model · Q 9.4 · Speed 9.3 Alibaba Cloud (Qwen Team) Qwen3-ASR Local ASR model · Q 9.5 · Speed 9
CompareBrowse all TTS models Local AIBrowse LLM models macOS appGet LocalClaw