Downloads · 30 days
11
30% of all-time downloads
trelis-mlx-hinglish-q4
sanke/trelis-mlx-hinglish-q4
trelis-mlx-hinglish-q4 is a automatic speech recognition model from sanke. Use it when you need speech turned into text. It is set up for mlx. The card lists the license as cc-by-nc-4.0.
This is a 4-bit MLX quantized version of the Trelis/whisper-hinglish-preview model by Trelis Research.
All-time downloads
37
Public
Repo size
878 MB
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors878 MB · 100%
At a glance
- Task
- Automatic Speech Recognition
- Library
- mlx
- License
- cc-by-nc-4.0
- Model type
- whisper
- Access
- Public
- Created
- Jul 22, 2026
- Updated
- Jul 22, 2026
- SHA
- ca667d8f
Model card
From the Hugging Face model README
Trelis Whisper Hinglish (Preview) - MLX Q4
This is a 4-bit MLX quantized version of the Trelis/whisper-hinglish-preview model by Trelis Research.
Conversion Details
The original model (a fine-tune of whisper-large-v3 for English/Hindi code-switching) has been unsharded and converted to the MLX format with 4-bit quantization. This reduces the memory footprint from ~3.1 GB down to ~733 MB, allowing it to run extremely fast on Apple Silicon (M-series) Macs while maintaining high accuracy.
Usage with MLX
You can load and run this model using the mlx-whisper package:
import mlx_whisper
text = mlx_whisper.transcribe(
"audio.wav",
path_or_hf_repo="sanke/trelis-mlx-hinglish-q4"
)
print(text["text"])
Credits & Links
- Original Model: Trelis/whisper-hinglish-preview
- Original Author: Trelis Research
- Framework: MLX (ml-explore)
- Task
- Automatic Speech Recognition
- Library
- mlx
- Type
- whisper
- License
- cc-by-nc-4.0
- Languages
- hi, en
- Created
- Jul 22, 2026
- Updated
- Jul 22, 2026