I'm trying to use Azure Speech to text service. In the documentation I'm confronted with examples, that use V1 API version: https://$region.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1

And basically every link to proper documentation is for the V3 API.

https://{endpoint}/speechtotext/v3.0

In this V1 example you can easily send your file as binary.

curl --location --request POST \
"https://$region.stt.speech.microsoft.com/speech/recognition/conversation/cognitiveservices/v1?language=en-US" \
--header "Ocp-Apim-Subscription-Key: $key" \
--header "Content-Type: audio/wav" \
--data-binary $audio_file

But I could not figure it out how to provide an wordLevelTimestampsEnabled=true parameter for getting word level timestamps.

On the other hand, I tried using the V3 API, and I can easily provide wordLevelTimestampsEnabled=true parameter, but I couldn't figure out how to send binary file data.

curl -L -X POST 'https://northeurope.api.cognitive.microsoft.com/speechtotext/v3.0/transcriptions' -H 'Content-Type: application/json' -H 'Accept: application/json' -H 'Ocp-Apim-Subscription-Key: $key' --data-raw '{
  "contentUrls": [
    "https://url-to-file.dev/test-file.wav"
  ],
  "properties": {
    "diarizationEnabled": false,
    "wordLevelTimestampsEnabled": true,
    "punctuationMode": "DictatedAndAutomatic",
    "profanityFilterMode": "Masked"
  },
  "locale": "pl-PL",
  "displayName": "Transcription using default model for pl-PL"
}'

Is there a way to pass a binary file and also get word level timestamps with wordLevelTimestampsEnabled=true parameter?

CodePudding user response：

Is there a way to pass a binary file and also get word level timestamps with wordLevelTimestampsEnabled=true parameter?

As suggested by Code Different, converting a comment as a community wiki answer to help community members who might face a similar issue.

As per the documentation, binary file can't be uploaded directly. You should provide URL via contentUrls property.

For example:

{
  "contentUrls": [
    "<URL to an audio file to transcribe>",
  ],
  "properties": {
    "diarizationEnabled": false,
    "wordLevelTimestampsEnabled": true,
    "punctuationMode": "DictatedAndAutomatic",
    "profanityFilterMode": "Masked"
  },
  "locale": "en-US",
  "displayName": "Transcription of file using default model for en-US"
}

You can refer to Speech-to-text REST API v3.0, cognitive-services-speech-sdk and Azure Speech Recognition - use binary / hexadecimal data instead of WAV file path