Contact UsGet started

Synchronous speech recognition using the Python SDK

Written by
Updated at February 16, 2024

Below, we provide an example of synchronous speech recognition from an audio file using the SpeechKit Python SDK. This example uses the following parameters:

To use the Python SDK, the yandex-speechkit package is required.

Authentication is performed under a service account using an API key or IAM token. Learn more about authentication in the SpeechKit API.

Getting started

  1. Create a service account and assign it the ai.speechkit-stt.user role.
  2. Get an API key for the service account and save it.
  3. Download a sample audio file for recognition or generate your own one.

Create an application for synchronous speech recognition

  1. Install the yandex-speechkit package using the pip package manager:

    pip install yandex-speechkit

    The installation was tested on Python 3.9. For the minimum allowed Python version, see the SDK website.

    If a grpcio-tools package version conflict occurs, see Resolving version conflicts during the installation of Python SDK.

  2. Create a file named test.py and add the following code to it:

    from argparse import ArgumentParser
from speechkit import model_repository, configure_credentials, creds
from speechkit.stt import AudioProcessingType

# Authentication via an API key.
configure_credentials(
   yandex_credentials=creds.YandexCredentials(
      api_key='<API key>'
   )
)

def recognize(audio):
   model = model_repository.recognition_model()

   # Set the recognition settings.
   model.model = 'general'
   model.language = 'ru-RU'
   model.audio_processing_type = AudioProcessingType.Full

   # Recognition of speech in the specified audio file and output of results to the console.
   result = model.transcribe_file(audio)
   for c, res in enumerate(result):
      print('=' * 80)
      print(f'channel: {c}\n\nraw_text:\n{res.raw_text}\n\nnorm_text:\n{res.normalized_text}\n')
      if res.has_utterances():
         print('utterances:')
         for utterance in res.utterances:
            print(utterance)

if __name__ == '__main__':
   parser = ArgumentParser()
   parser.add_argument('--audio', type=str, help='audio path', required=True)

   args = parser.parse_args()

   recognize(args.audio)

    Where:

    • api_key: API key for the service account.

    • audio: Path to the file for audio recording.

    • model: Recognition model.

    • language: Recognition language.

    • audio_processing_type: Audio processing method.

      The Python SDK does not support streaming and asynchronous recognition, but you can simulate these features. To do this, set the following value in the test.py file, the audio_processing_type parameter:

      • AudioProcessingType.Stream for streaming recognition.
      • AudioProcessingType.Full for asynchronous recognition.

  3. Run the created file:

    python3 test.py --audio speech.pcm

    Where --audio is the path to the audio file to transcribe.

    The result contains recognized speech:

    channel: 0

raw_text:
i'm yandex speechkit i can turn any text into speech now you can too

norm_text:
I'm Yandex SpeechKit. I can turn any text into speech. Now you can, too!

utterances:
I'm Yandex SpeechKit. I can turn any text into speech. Now you can, too! [0.419, 6.379]

See also

Previous
Authentication in the Python SDK
Next
Speech synthesis using the Python SDK