Downloads · 30 days
37
19% of all-time downloads
MOSS-Transcribe-Diarize-MLX-8bit
kuotient/MOSS-Transcribe-Diarize-MLX-8bit
MOSS-Transcribe-Diarize-MLX-8bit is a automatic speech recognition model from kuotient. Use it when you need speech turned into text. It is set up for mlx-audio. The card lists the license as apache-2.0.
8-bit MLX quantization of OpenMOSS-Team/MOSS-Transcribe-Diarize (0.9B end-to-end transcription + speaker diarization + timestamps).
All-time downloads
195
Public
Parameters
909M
2.7 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors1.4 GB · 99%
Parameter types
How the weights are stored.
BF16468M · 52%
Base models
Model card
From the Hugging Face model README
MOSS-Transcribe-Diarize — MLX 8-bit
8-bit MLX quantization of OpenMOSS-Team/MOSS-Transcribe-Diarize (0.9B end-to-end transcription + speaker diarization + timestamps).
- Quantization: affine 8-bit, group size 64, text backbone only — the Whisper encoder and the VQ adaptor stay full precision (quantizing the encoder's positional embedding breaks the feature broadcast).
- Quality: byte-identical transcripts to the bf16 reference across Korean, Korean↔English code-switching, and multi-speaker English in our checks.
- Load:
mlx-audio/mlx-audio-swiftviafromModelDirectory.
Original model © the OpenMOSS / MOSI.AI team, Apache-2.0. This repository only re-hosts a quantized copy for reproducible on-device deployment.
- Task
- Automatic Speech Recognition
- Library
- mlx-audio
- Type
- moss_transcribe_diarize
- License
- apache-2.0
- Created
- Jul 23, 2026
- Updated
- Jul 24, 2026