hearing.backends.mlx_whisper.adapter¶

Adapter for the mlx-whisper backend (Whisper on Apple Silicon via MLX).

Decodes the input to a float32 mono 16 kHz array (no ffmpeg needed), calls mlx_whisper.transcribe(array, path_or_hf_repo=...) and maps its openai-whisper-shaped result dict onto a scribed.base.Transcript. mlx_whisper is imported on first use, after the Apple-Silicon check, so this module imports anywhere.

Module Attributes

`MODEL_REPO_TEMPLATE`	How a bare Whisper size maps to an `mlx-community` repo id.
`MODEL_REPO_OVERRIDES`	Sizes whose mlx-community repo does not follow `MODEL_REPO_TEMPLATE`.

Functions

resolve_model_repo(model)

Resolve a model name to a Hugging Face repo id for mlx_whisper.

Classes

Adapter(config)

mlx-whisper adapter (Apple Silicon only).

class hearing.backends.mlx_whisper.adapter.Adapter(config)[source]¶

Bases: BaseTranscriberAdapter

mlx-whisper adapter (Apple Silicon only).

module_name = 'mlx_whisper'¶: Engine module name — the seam tests use to inject a fake backend.

hearing.backends.mlx_whisper.adapter.MODEL_REPO_OVERRIDES = {'large-v3-turbo': 'mlx-community/whisper-large-v3-turbo', 'turbo': 'mlx-community/whisper-large-v3-turbo'}¶: Sizes whose mlx-community repo does not follow MODEL_REPO_TEMPLATE.

hearing.backends.mlx_whisper.adapter.MODEL_REPO_TEMPLATE = 'mlx-community/whisper-{model}-mlx'¶: How a bare Whisper size maps to an mlx-community repo id.

hearing.backends.mlx_whisper.adapter.resolve_model_repo(model)[source]¶

Resolve a model name to a Hugging Face repo id for mlx_whisper.

Return type:: str

>>> resolve_model_repo("base")
'mlx-community/whisper-base-mlx'
>>> resolve_model_repo("turbo")
'mlx-community/whisper-large-v3-turbo'
>>> resolve_model_repo("someone/custom-whisper")
'someone/custom-whisper'