Activer le filtre de langage grossier

Cette page explique comment utiliser Speech-to-Text pour détecter automatiquement les mots grossiers dans vos données audio et les censurer dans la transcription.

Vous pouvez activer le filtre contre les grossièretés en définissant profanityFilter=true dans RecognitionConfig. Si cette option est activée, Speech-to-Text tente de détecter les mots grossiers et de ne renvoyer que la première lettre suivie d'astérisques dans la transcription (par exemple, f***). Si ce champ est défini sur false ou n'est pas défini, Speech-to-Text ne tente pas de filtrer les grossièretés.

L'exemple suivant montre comment activer le filtre contre les grossièretés pour détecter des contenus audio stockés dans un bucket Google Cloud Storage.

Java

import com.google.cloud.speech.v1.RecognitionAudio;
import com.google.cloud.speech.v1.RecognitionConfig;
import com.google.cloud.speech.v1.RecognitionConfig.AudioEncoding;
import com.google.cloud.speech.v1.RecognizeResponse;
import com.google.cloud.speech.v1.SpeechClient;
import com.google.cloud.speech.v1.SpeechRecognitionAlternative;
import com.google.cloud.speech.v1.SpeechRecognitionResult;
import java.util.List;

public class SpeechProfanityFilter {

  public void speechProfanityFilter() throws Exception {
    String uriPath = "gs://cloud-samples-tests/speech/brooklyn.flac";
    speechProfanityFilter(uriPath);
  }

  /**
   * Transcribe a remote audio file with multi-channel recognition
   *
   * @param gcsUri the path to the audio file
   */
  public static void speechProfanityFilter(String gcsUri) throws Exception {
    // Instantiates a client with GOOGLE_APPLICATION_CREDENTIALS
    try (SpeechClient speech = SpeechClient.create()) {

      // Configure remote file request
      RecognitionConfig config =
          RecognitionConfig.newBuilder()
              .setEncoding(AudioEncoding.FLAC)
              .setLanguageCode("en-US")
              .setSampleRateHertz(16000)
              .setProfanityFilter(true)
              .build();

      // Set the remote path for the audio file
      RecognitionAudio audio = RecognitionAudio.newBuilder().setUri(gcsUri).build();

      // Use blocking call to get audio transcript
      RecognizeResponse response = speech.recognize(config, audio);
      List<SpeechRecognitionResult> results = response.getResultsList();

      for (SpeechRecognitionResult result : results) {
        // There can be several alternative transcripts for a given chunk of speech. Just use the
        // first (most likely) one here.
        SpeechRecognitionAlternative alternative = result.getAlternativesList().get(0);
        System.out.printf("Transcription: %s\n", alternative.getTranscript());
      }
    }
  }
}

Node.js

// Filters profanity

/**
 * TODO(developer): Uncomment these variables before running the sample.
 */
// const gcsUri = 'gs://my-bucket/audio.raw';

async function syncRecognizeWithProfanityFilter() {
  // Imports the Google Cloud client library
  const speech = require('@google-cloud/speech');

  // Creates a client
  const client = new speech.SpeechClient();

  const audio = {
    uri: gcsUri,
  };

  const config = {
    encoding: 'FLAC',
    sampleRateHertz: 16000,
    languageCode: 'en-US',
    profanityFilter: true, // set this to true
  };
  const request = {
    audio: audio,
    config: config,
  };

  // Detects speech in the audio file
  const [response] = await client.recognize(request);
  const transcription = response.results
    .map(result => result.alternatives[0].transcript)
    .join('\n');
  console.log(`Transcription: ${transcription}`);
}
syncRecognizeWithProfanityFilter().catch(console.error);

Python

def sync_recognize_with_profanity_filter_gcs(gcs_uri):

    from google.cloud import speech

    client = speech.SpeechClient()

    audio = {"uri": gcs_uri}

    config = speech.RecognitionConfig(
        encoding=speech.RecognitionConfig.AudioEncoding.FLAC,
        sample_rate_hertz=16000,
        language_code="en-US",
        profanity_filter=True,
    )

    response = client.recognize(config=config, audio=audio)

    for i, result in enumerate(response.results):
        alternative = result.alternatives[0]
        print(u"Transcript: {}".format(alternative.transcript))