LocalWhisperTranscriber

Transcribe audio files using OpenAI's Whisper model running locally on your machine.

Key Features

• Transcribes audio to text using a locally downloaded Whisper model.
• Supports multiple model sizes from tiny to large-v3.
• Accepts audio files as file paths, Path objects, or ByteStream objects.
• Returns transcriptions as Haystack Document objects.
• Configurable device support (CPU, GPU) for inference.
• Supports additional Whisper parameters like language, task, and temperature.

Configuration

1. Drag the LocalWhisperTranscriber component onto the canvas from the Component Library.
2. Click on the component to open the configuration panel.
3. On the General tab:
   1. Set the model to the Whisper model size you want to use. Larger models are more accurate but slower. Available sizes are: tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1, large-v2, and large-v3.
4. Go to the Advanced tab to configure device and whisper_params for options like language and task.

Connections

LocalWhisperTranscriber receives a list of audio file sources. It outputs a list of Document objects containing the transcription text. Connect its documents output to a splitter, embedder, or other pipeline components.

Source Code

To check this component's source code, open whisper_local.py in the Haystack Core Integrations repository.

Usage Examples

Basic Configuration

  LocalWhisperTranscriber:
    type: haystack_integrations.components.audio.whisper.LocalWhisperTranscriber
    init_parameters:
      model: large
      whisper_params:
        language: en

Using the Component in a Pipeline

# haystack-pipeline
components:
  LocalWhisperTranscriber:
    type: haystack_integrations.components.audio.whisper.LocalWhisperTranscriber
    init_parameters:
      model: base
      whisper_params:
        language: en
        task: transcribe

  document_splitter:
    type: haystack.components.preprocessors.document_splitter.DocumentSplitter
    init_parameters:
      split_by: word
      split_length: 250
      split_overlap: 30

connections:
- sender: LocalWhisperTranscriber.documents
  receiver: document_splitter.documents

max_runs_per_component: 100

metadata: {}

Parameters

Inputs

Parameter	Type	Description
sources	List[Union[str, Path, ByteStream]]	A list of audio file paths, Path objects, or ByteStream objects to transcribe.
whisper_params	Optional[Dict[str, Any]]	Additional parameters for the Whisper transcription, overriding init-time values.

Outputs

Parameter	Type	Description
documents	List[Document]	The transcriptions as Haystack Documents.

Init Parameters

These are the parameters you can configure in Pipeline Builder:

Parameter	Type	Default	Description
model	str	large	The Whisper model size to use. Options are tiny, tiny.en, base, base.en, small, small.en, medium, medium.en, large, large-v1, large-v2, and large-v3. The .en variants are English-only models.
device	Optional[ComponentDevice]	None	The device to run inference on. If not set, automatically detected.
whisper_params	Optional[Dict[str, Any]]	None	Additional parameters passed to whisper.transcribe(). Supports language (ISO 639-1 code), task ("transcribe" or "translate"), temperature, and other Whisper options.

Run Method Parameters

These are the parameters you can configure for the component's run() method. This means you can pass these parameters at query time through the API, in Playground, or when running a job. For details, see Modify Pipeline Parameters at Query Time.

Parameter	Type	Default	Description
sources	List[Union[str, Path, ByteStream]]		A list of audio file paths or streams to transcribe.
whisper_params	Optional[Dict[str, Any]]	None	Additional parameters to override init-time values.

Related Information

• Pipeline Components