MOSS Transcribe Diarize

MOSS-Transcribe-Diarize 0.9B is an end-to-end audio model that transcribes speech, labels speakers, and adds timestamps in a single pass, turning long, multi-speaker recordings into structured "who said what."

It supports 50+ languages, single-pass inference on recordings up to 90 minutes, acoustic-event awareness, and custom hotword prompting for domain-specific terms.

Performance

The charts below show median latency and unit economics (cost per audio hour) change with audio length when transcribing prerecorded files at 1, 4 and 16 concurrent requests per GPU. Higher concurrency significantly reduces cost per audio hour, with a corresponding tradeoff in latency that grows with audio length.

Concurrency is configurable and can be tuned to achieve the latency and cost targets that best fit your workload and business requirements. See the concurrency guide for configuration details.

For batch workloads, we recommend starting with H100 using concurrency 16, and lowering it to reduce latency on longer audio or raising it to increase throughput on latency tolerant workloads

✕

Batch transcription latency and cost across audio lengths at concurrency 1, 4 and 16. Running more requests in parallel lowers cost per audio hour, while P50 latency grows with audio length, most sharply at concurrency 16.

Using the model

It serves the OpenAI-compatible /v1/audio/transcriptions endpoint, and diarization and timestamps are on by default. Request verbose_json to get parsed speaker segments.