Transcribe audio from streaming input

This section demonstrates how to transcribe streaming audio, like the input from a microphone, to text.

Streaming speech recognition allows you to stream audio to Speech-to-Text and receive a stream speech recognition results in real time as the audio is processed. See also the audio limits for streaming speech recognition requests. Streaming speech recognition is available via gRPC only.

For more information about recognizers and sending recognition requests, see the reference documentation.

Note: To use streaming recognition to stop listening after the user speaks a single word, like in the case of voice commands, use the latest_short model in your Recognizer. The latest_short model tells the Speech-to-Text to end the transcription request once it detects that the speech has ended like at the end of a single word or phrase.

Perform streaming speech recognition on a local file

Below is an example of performing streaming speech recognition on a local audio file. There is a 10 MB limit on all streaming requests sent to the API. This limit applies to to both the initial StreamingRecognize request and the size of each individual message in the stream. Exceeding this limit will throw an error.

Python

To authenticate to Speech-to-Text, set up Application Default Credentials. For more information, see Set up authentication for a local development environment.

View on GitHub Feedback

from google.cloud.speech_v2 import SpeechClient
from google.cloud.speech_v2.types import cloud_speech


def transcribe_streaming_v2(
        project_id: str,
        recognizer_id: str,
        audio_file: str,
) -> cloud_speech.StreamingRecognizeResponse:
    """Transcribes audio from audio file stream.

    Args:
        project_id: The GCP project ID.
        recognizer_id: The ID of the recognizer to use.
        audio_file: The path to the audio file to transcribe.

    Returns:
        The response from the transcribe method.
    """
    # Instantiates a client
    client = SpeechClient()

    request = cloud_speech.CreateRecognizerRequest(
        parent=f"projects/{project_id}/locations/global",
        recognizer_id=recognizer_id,
        recognizer=cloud_speech.Recognizer(
            language_codes=["en-US"], model="latest_long"
        ),
    )

    # Creates a Recognizer
    operation = client.create_recognizer(request=request)
    recognizer = operation.result()

    # Reads a file as bytes
    with open(audio_file, "rb") as f:
        content = f.read()

    # In practice, stream should be a generator yielding chunks of audio data
    chunk_length = len(content) // 5
    stream = [
        content[start : start + chunk_length]
        for start in range(0, len(content), chunk_length)
    ]
    audio_requests = (
        cloud_speech.StreamingRecognizeRequest(audio=audio) for audio in stream
    )

    recognition_config = cloud_speech.RecognitionConfig(auto_decoding_config={})
    streaming_config = cloud_speech.StreamingRecognitionConfig(
        config=recognition_config
    )
    config_request = cloud_speech.StreamingRecognizeRequest(
        recognizer=recognizer.name, streaming_config=streaming_config
    )

    def requests(
            config: cloud_speech.RecognitionConfig, audio: list
    ) -> list:
        yield config
        yield from audio

    # Transcribes the audio into text
    responses_iterator = client.streaming_recognize(
        requests=requests(config_request, audio_requests)
    )
    responses = []
    for response in responses_iterator:
        responses.append(response)
        for result in response.results:
            print(f"Transcript: {result.alternatives[0].transcript}")

    return responses

While you can stream a local audio file to the Speech-to-Text API, it is recommended that you perform synchronous audio recognition.

Try it for yourself

If you're new to Google Cloud, create an account to evaluate how Speech-to-Text performs in real-world scenarios. New customers also get $300 in free credits to run, test, and deploy workloads.

Try Speech-to-Text free