README.md

metadata

license: apache-2.0
library_name: transformers
language:
  - en
  - zh
tags:
  - moss
  - audio
  - speech
  - asr
  - diarization
  - timestamp-asr
  - long-form-audio
  - multimodal
  - multilingual
  - custom_code
pipeline_tag: audio-text-to-text

MOSS-Transcribe-Diarize

    

MOSS-Transcribe-Diarize 0.9B is an open-source SOTA end-to-end audio understanding model for long-form multi-speaker transcription, diarization, timestamps, and acoustic event awareness. MOSS-Transcribe-Diarize Pro is a stronger model with higher overall performance and will be available through API access soon.

News

• 2026-07-14: 🏆 MOSS-Transcribe-Diarize won first place in the 2nd MLC-SLM Challenge at INTERSPEECH 2026, covering 14 languages.
• 2026-07-09: Open-sourced MOSS-Transcribe-Diarize 0.9B.

Contents

• Introduction
• Model Architecture
• Evaluation
  • Objective Evaluation
• Quickstart
  • Environment Setup
  • Python Usage
  • Serve with SGLang Omni
  • Serve with vLLM
  • Custom Prompt and Hotwords
  • Subtitle Web App
• Citation

Introduction

MOSS-Transcribe-Diarize is our flagship SOTA model family for turning real-world long-form audio into structured, speaker-aware transcripts in one pass. Instead of stitching together separate ASR and diarization systems, these models jointly perform speech transcription and speaker diarization, producing time-aligned text with precise timestamps and consistent speaker labels such as [S01], [S02], and beyond.

Built for meetings, calls, podcasts, interviews, lectures, and video content, MOSS-Transcribe-Diarize is designed to handle long, messy, multi-speaker recordings where reliability matters. It can also emit optional acoustic event annotations, giving downstream systems a richer understanding of what happened, who spoke, and when.

MOSS-Transcribe-Diarize supports 50+ languages.

The model accepts raw audio and emits a compact timestamped transcript. The canonical output format is:

[start_time][Sxx]transcribed speech[end_time]

Timestamps are expressed in seconds, and adjacent segments are concatenated into a single stream, for example:

[0.48][S01]Welcome everyone[1.66][12.26][S02]The new transcription pipeline is ready for evaluation[13.81][14.36][S01]Great, include the diarization results in the report[18.76]

Model Architecture

Component	Specification
Text backbone	Qwen3-0.6B style causal decoder
Audio encoder	Whisper-Medium encoder configuration
Audio frontend	WhisperFeatureExtractor, 16 kHz, 80 mel bins, 30 s chunks
Audio-text bridge	4x temporal merge + MLP adaptor
Fusion	Audio features replace <|audio_pad|> embeddings via masked_scatter
Output format	Compact [start][Sxx]text[end] transcript with speaker tags such as [S01]

This Hugging Face repository includes the custom Transformers remote code required to load the model with trust_remote_code=True.

Evaluation

Objective Evaluation

We evaluate MOSS-Transcribe-Diarize using three objective metrics: Character Error Rate (CER), concatenated minimum-permutation Character Error Rate (cpCER), and Δcp. Lower is better for all metrics. Best results are bolded, second-best results are underlined. A dash (-) indicates that the result is unavailable.

Model	AISHELL‑4			Alimeeting			Podcast			Movies
	CER↓	cpCER↓	Δcp↓	CER↓	cpCER↓	Δcp↓	CER↓	cpCER↓	Δcp↓	CER↓	cpCER↓	Δcp↓
Doubao	18.18	27.86	9.68	25.25	37.57	12.31	7.93	10.54	2.61	9.94	30.88	20.94
ElevenLabs	19.58	37.95	18.36	25.70	36.69	10.99	8.50	11.34	2.85	11.49	17.85	6.37
GPT-4o	-	-	-	-	-	-	-	-	-	14.37	23.67	9.31
Gemini 2.5 Pro	42.70	53.42	10.72	27.43	41.64	14.21	7.38	10.23	2.85	15.46	24.15	8.69
Gemini 3 Pro	22.75	27.43	4.68	26.75	32.84	6.09	-	-	-	8.62	14.73	6.11
VIBEVOICE ASR	21.40	24.99	3.59	27.40	29.33	1.93	27.94	48.30	20.36	14.59	42.54	27.94
MOSS Transcribe Diarize 0.9B	14.84	15.83	0.99	24.86	22.17	-2.69	5.97	7.37	1.40	6.36	12.76	6.40
MOSS Transcribe Diarize Pro	13.78	14.02	0.24	18.22	13.94	-4.27	4.46	6.97	2.51	5.86	11.78	5.92

Quickstart

Environment Setup

Use a clean Python environment. The project is tested with Python 3.12 and Transformers 5.x.

git clone https://github.com/OpenMOSS/MOSS-Transcribe-Diarize.git
cd MOSS-Transcribe-Diarize
uv venv --python 3.12 .venv
source .venv/bin/activate
uv pip install -e ".[torch-runtime]" --torch-backend=auto

The GitHub package provides helper utilities such as audio/video loading, transcription message construction, transcript parsing, CLI inference, and the subtitle web app. The model weights and remote-code model files are loaded from this Hugging Face repository.

Python Usage

import torch
from transformers import AutoModelForCausalLM, AutoProcessor

from moss_transcribe_diarize import parse_transcript
from moss_transcribe_diarize.inference_utils import (
    build_transcription_messages,
    generate_transcription,
    resolve_device,
)

model_id = "OpenMOSS-Team/MOSS-Transcribe-Diarize"
audio_path = "audio.wav"

device = resolve_device("auto")
dtype = torch.bfloat16 if device.type == "cuda" else torch.float32

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    trust_remote_code=True,
    dtype="auto",
).to(dtype=dtype).to(device).eval()
processor = AutoProcessor.from_pretrained(model_id, trust_remote_code=True)

messages = build_transcription_messages(audio_path)
result = generate_transcription(
    model,
    processor,
    messages,
    max_new_tokens=2048,
    do_sample=False,
    device=device,
    dtype=dtype,
)

print(result["text"])

for segment in parse_transcript(result["text"]):
    print(segment.start, segment.end, segment.speaker, segment.text)

The message flow follows the common Qwen multimodal pattern. The chat template is loaded from the model by AutoProcessor:

1. processor.apply_chat_template(messages, tokenize=False) renders text with audio placeholders.
2. process_audio_info(messages, sampling_rate) loads audio waveforms from the same messages.
3. processor(text=text, audio=audios) computes Whisper input features and expands audio placeholders.
4. model.generate(...) produces timestamped transcription and diarization text.

Serve with SGLang Omni

SGLang Omni is the recommended serving backend for MOSS-Transcribe-Diarize, providing optimized long-form audio inference through the OpenAI-compatible /v1/audio/transcriptions endpoint.

SGLang Omni currently targets CUDA 13 environments. Please follow the official installation guide for the supported setup. For CUDA 12 environments, the vLLM workflow is also available below.

Download the model:

hf download OpenMOSS-Team/MOSS-Transcribe-Diarize

Serve the model:

sgl-omni serve \
  --model-path OpenMOSS-Team/MOSS-Transcribe-Diarize \
  --port 8000 \
  --max-running-requests 16 \
  --cuda-graph-max-bs 16 \
  --mem-fraction-static 0.80

Use response_format=verbose_json when you need parsed speaker segments. json returns the raw transcript text only.

curl -X POST http://localhost:8000/v1/audio/transcriptions \
  -F model=OpenMOSS-Team/MOSS-Transcribe-Diarize \
  -F file=@audio.wav \
  -F response_format=verbose_json

import requests

with open("audio.wav", "rb") as f:
    resp = requests.post(
        "http://localhost:8000/v1/audio/transcriptions",
        data={
            "model": "OpenMOSS-Team/MOSS-Transcribe-Diarize",
            "response_format": "verbose_json",
        },
        files={"file": ("audio.wav", f, "audio/wav")},
        timeout=300,
    )

resp.raise_for_status()
payload = resp.json()
print(payload["text"])
for segment in payload.get("segments", []):
    print(f"[{segment['start']:.2f}-{segment['end']:.2f}] {segment['text']}")

For longer multi-speaker audio, raise max_new_tokens so the decoder can finish the full diarized transcript:

curl -X POST http://localhost:8000/v1/audio/transcriptions \
  -F model=OpenMOSS-Team/MOSS-Transcribe-Diarize \
  -F file=@audio.wav \
  -F response_format=verbose_json \
  -F max_new_tokens=65536

Parameter	Type	Default	Description
file	file	required	Audio file uploaded as multipart form data
model	string	server default	Model identifier
language	string	unset	Optional language hint
response_format	string	json	json, verbose_json, or text
temperature	float	model default (0.0)	Sampling temperature
max_new_tokens	int	5120	Max generated tokens; raise for long audio, for example 65536
prompt	string	unset	Optional instruction override; omit to use the built-in transcribe+diarize prompt

For benchmarking, performance numbers, and implementation details, see the SGLang Omni cookbook. The following single-H100 results are reported for short- and long-sequence multi-speaker ASR tasks.

movies short-sequence ASR:

Concurrency	Throughput (req/s)	Mean latency (s)	RTF mean	audio_s/s
1	2.57	0.388	0.0612	29.76
2	4.89	0.409	0.0659	56.55
4	6.62	0.513	0.0790	76.64
8	6.80	0.533	0.0810	78.70
16	7.08	0.659	0.0922	81.98

aishell4_long long-sequence ASR:

Concurrency	Throughput (req/s)	Mean latency (s)	RTF mean	audio_s/s
1	0.022	45.2	0.0197	50.64
2	0.032	60.7	0.0265	74.25
4	0.036	105.6	0.0461	81.64
8	0.040	172.6	0.0754	90.62
16	0.043	282.8	0.1237	98.83

Serve with vLLM

MOSS-Transcribe-Diarize supports vLLM serving through the OpenAI-compatible transcription API. Use a pinned vLLM nightly build that includes the MOSS-Transcribe-Diarize model registration. Choose one of the following commands: for CUDA 12 environments, use cu129; for CUDA 13 environments, use cu130.

uv pip install -U vllm \
  --torch-backend=auto \
  --extra-index-url https://wheels.vllm.ai/68b4a1d582818e67adc903bf1b8fc5a5447da2fa/cu129

or:

uv pip install -U vllm \
  --torch-backend=auto \
  --extra-index-url https://wheels.vllm.ai/68b4a1d582818e67adc903bf1b8fc5a5447da2fa/cu130

vllm serve OpenMOSS-Team/MOSS-Transcribe-Diarize --trust-remote-code

curl http://localhost:8000/v1/audio/transcriptions \
  -F model="OpenMOSS-Team/MOSS-Transcribe-Diarize" \
  -F file=@"audio.wav" \
  -F response_format="json" \
  -F temperature="0"

Custom Prompt and Hotwords

The default prompt is optimized for timestamped transcription and speaker diarization:

请将音频转写为文本，每一段需以起始时间戳和说话人编号（[S01]、[S02]、[S03]…）开头，正文为对应的语音内容，并在段末标注结束时间戳，以清晰标明该段语音范围。

To add hotwords, append a short hint to the default prompt:

请将音频转写为文本，每一段需以起始时间戳和说话人编号（[S01]、[S02]、[S03]…）开头，正文为对应的语音内容，并在段末标注结束时间戳，以清晰标明该段语音范围。热词提示：热词1, 热词2, 热词3

More prompt recipes are available in examples/prompts.md. The same prompt can be passed to build_transcription_messages, mtd-subtitle, and mtd-subtitle-web.

Subtitle Web App

The package also includes a local subtitle workflow for upload, review, subtitle export, and optional FFmpeg burn-in:

mtd-subtitle-web \
  --model OpenMOSS-Team/MOSS-Transcribe-Diarize \
  --host 127.0.0.1 \
  --port 7860

Open http://127.0.0.1:7860, upload an audio/video file, review the parsed subtitle segments, then download JSON/SRT/ASS or burn an MP4 if ffmpeg and ffprobe are available on PATH.

For batch processing:

mtd-subtitle /path/to/input.mp4 \
  --model OpenMOSS-Team/MOSS-Transcribe-Diarize \
  --out-dir runs/example \
  --render

Citation

If you use MOSS-Transcribe-Diarize, please cite the technical report:

@misc{moss_transcribe_diarize_2026,
  title={MOSS Transcribe Diarize Technical Report},
  author={{MOSI.AI}},
  year={2026},
  eprint={2601.01554},
  archivePrefix={arXiv},
  primaryClass={cs.SD},
  url={https://arxiv.org/abs/2601.01554}
}