Error with enable_speaker_diarization tag in Google Cloud Speech to Text

Question

Using Google-Speech-to-Text, I am able to transcribe an audio clip with the default parameters. However, I get an error message while using the enable_speaker_diarization tag to profile individual speakers in the audio clip. Google documents it here This is a long recognize audio clip hence I am using async request which Google recommends here

My code -

def transcribe_gcs(gcs_uri):
from google.cloud import speech
from google.cloud import speech_v1 as speech
from google.cloud.speech import enums
from google.cloud.speech import types
client = speech.SpeechClient()
audio = types.RecognitionAudio(uri = gcs_uri)
config = speech.types.RecognitionConfig(encoding=speech.enums.RecognitionConfig.AudioEncoding.FLAC, 
                                        sample_rate_hertz= 16000, 
                                        language_code = 'en-US',
                                       enable_speaker_diarization=True,
                                        diarization_speaker_count=2)

operation = client.long_running_recognize(config, audio)
print('Waiting for operation to complete...')
response = operation.result(timeout=3000)
result = response.results[-1]

words_info = result.alternatives[0].words

for word_info in words_info:
    print("word: '{}', speaker_tag: {}".format(word_info.word, word_info.speaker_tag))

After using -

transcribe_gcs('gs://bucket_name/filename.flac') 

I get the error

ValueError: Protocol message RecognitionConfig has no "enable_speaker_diarization" field.

I am sure this is something to do with libraries, I have used all variants I could find like

from google.cloud import speech_v1p1beta1 as speech
from google.cloud import speech

But I keep getting the same error. Note - I have already authenticated using the JSON file prior to running this code.

Answer 1

The enable_speaker_diarization=True parameter in speech.types.RecognitionConfig is available only in the library speech_v1p1beta1 at the moment, so, you need to import that library in order to use that parameter, not the default speech one. I did some modifications to your code and works fine for me. Take into account that you need to use a service account to run this code.

def transcribe_gcs(gcs_uri):
    from google.cloud import speech_v1p1beta1 as speech
    from google.cloud.speech_v1p1beta1 import enums
    from google.cloud.speech_v1p1beta1 import types
    client = speech.SpeechClient()
    audio = types.RecognitionAudio(uri = gcs_uri)
    config = speech.types.RecognitionConfig( language_code = 'en-US',enable_speaker_diarization=True, diarization_speaker_count=2)
    operation = client.long_running_recognize(config, audio)
    print('Waiting for operation to complete...')
    response = operation.result(timeout=3000)
    result = response.results[-1]

    words_info = result.alternatives[0].words
    
    tag=1
    speaker=""

    for word_info in words_info:
        if word_info.speaker_tag==tag:
            speaker=speaker+" "+word_info.word

        else:
            print("speaker {}: {}".format(tag,speaker))
            tag=word_info.speaker_tag
            speaker=""+word_info.word

     print("speaker {}: {}".format(tag,speaker))

And the result should be like:

Answer 2

The error is because you have not imported some files. To do so, import the following files.

from google.cloud import speech_v1p1beta1 as speech
from google.cloud.speech_v1p1beta1 import enums
from google.cloud.speech_v1p1beta1 import types

Answer 3

The error cause is similar to Node JS users as well. Import the beta feature via this call, and then use the speaker identification features.

const speech = require('@google-cloud/speech').v1p1beta1;