README.md

metadata

license: apache-2.0
language:
  - de
library_name: mlx
pipeline_tag: automatic-speech-recognition
model-index:
  - name: >-
      mlx version of whisper-large-v3-turbo-german by Florian Zimmermeister
      @primeLine
    results:
      - task:
          type: automatic-speech-recognition
          name: Speech Recognition
        dataset:
          name: German ASR Data-Mix
          type: flozi00/asr-german-mixed
        metrics:
          - type: wer
            value: 2.628 %
            name: Test WER
datasets:
  - flozi00/asr-german-mixed
  - flozi00/asr-german-mixed-evals
base_model:
  - primeline/whisper-large-v3-german

whisper-large-v3-turbo-german-f16-q4

This model was converted to MLX format from primeline/whisper-large-v3-turbo-german and is quantized to 4bit, float16.

made with a custom script for converting safetensor whisper models.

there is also an unquantized float16 version

Use with MLX

git clone https://github.com/ml-explore/mlx-examples.git
cd mlx-examples/whisper/
pip install -r requirements.txt

import mlx_whisper
result = mlx_whisper.transcribe("test.mp3", path_or_hf_repo="mlx-community/whisper-large-v3-turbo-german-f16")
print(result)

whisper-large-v3-turbo-german-f16-q4

This model was converted to MLX format.

Use with MLX

git clone https://github.com/ml-explore/mlx-examples.git
cd mlx-examples/whisper/
pip install -r requirements.txt

# Example usage
import mlx_whisper
result = mlx_whisper.transcribe("test.mp3", path_or_hf_repo="whisper-large-v3-turbo-german-f16-q4")
print(result)