MOSS-Transcribe-Diarize
Joint transcription and in-decoder speaker diarization in one 0.9B-parameter model
Opens the desktop app and downloads this model for you.
Don't have the app yet? Download OpenASR Desktop
let's line up this week's release plan,
the beta build can ship on Friday,
then loop in design and QA,
Why it fits
MOSS-Transcribe-Diarize is a 0.9B-parameter joint speech-recognition and speaker-diarization
model from the OpenMOSS team. Its architecture pairs a Whisper-Medium-scale audio encoder with a
lightweight MLP/LayerNorm adapter feeding into a Qwen3-0.6B-scale decoder, enabling the model to
transcribe speech and attribute each segment to a recording-local speaker with start/end timestamps
in a single forward pass. This native speaker source replaces the external segmenter and clusterer,
but Voice ID naming still runs through the shared ReDimNet2-B6 evidence/matching stage. The catalog
lists Mandarin plus the 14 additional languages named by upstream. OpenASR distributes this model in
three quantization tiers -- fp16, q8_0, and q4_k -- packaged in the native .oasr runtime format for
CPU-based local inference.
What you can do with it
- 🎯 Transcription + anonymous speaker turns in one decode — one inference run emits text, recording-local speaker labels, and timestamps without the external segmenter/clusterer
- 🧬 Shared Voice ID after decode — naming an enrolled person still uses OpenASR's ReDimNet2-B6 evidence and identity stage; native diarization is not a persistent identity database
- 📊 Verified accuracy: 2.52% CER (Chinese), 2.23% WER (English) — benchmarked by OpenASR on frozen evaluation datasets, not upstream-reported numbers
- 🌐 15 cataloged languages — Mandarin plus 14 languages named by upstream, including English
- 📦 Three quantization tiers: fp16 / q8_0 / q4_k — delivered in OpenASR's native .oasr format for CPU inference, choose the precision-to-size tradeoff that fits your hardware
Other models
Developer details CLI commands, file hashes, and per-quant downloads — for scripting and verification.
Model facts
Usage
These are CLI / local-server examples. The desktop app runs this model without typing a command — use Install above.
$ openasr pull moss-transcribe-diarize:q8 ↓ moss-transcribe-diarize.oasr 1 GB ✓ verified sha256 $ openasr transcribe meeting.wav --backend native --model-pack ~/.openasr/models/moss-transcribe-diarize/q8_0/moss-transcribe-diarize-q8_0.oasr ✓ local transcript · 0 bytes sent
$ openasr serve --backend native --model-pack ~/.openasr/models/moss-transcribe-diarize/q8_0/moss-transcribe-diarize-q8_0.oasr --addr 127.0.0.1:8080 ▶ http://127.0.0.1:8080 · model=moss-transcribe-diarize · 0 bytes will leave this host
from openai import OpenAI client = OpenAI(base_url="http://127.0.0.1:8080/v1", api_key="local") audio = open("meeting.wav", "rb") text = client.audio.transcriptions.create(model="moss-transcribe-diarize", file=audio)