ASR Speech Recognition

Inputs

• input: Raw audio data containing the speech

Outputs

• output: Transcribed text from the input audio.

Configuration

• source: The path to a model or the name of a pre-trained model.
• device: The device to run the model on (e.g. “cpu” or “cuda”).
• chunk_size: The size of each chunk of audio data to process.
• left_context_size: The number of chunks to consider as context for each chunk.

Description

Receives audio data and transcribes it.

Notes

This task uses a pre-trained ASR model from SpeechBrain and transcribes the input audio in real-time. It is recommended to use a high-performance device (e.g. GPU) to run this task efficiently.