[nnnn].word.json file not found in results wordBoundaryEnabled: true

Motyar D 0

I am using Azure AI Batch Syntheses something like this
https://learn.microsoft.com/en-us/rest/api/batchtexttospeech/batch-syntheses/create?view=rest-batchtexttospeech-2024-04-01&tabs=HTTP#create-a-batch-synthesis-job-with-ssml-input-and-user-provided-container-output.

Everything is working well

But even with wordBoundaryEnabled: true
I can't find the [nnnn].word.json file.

I only get these three into output

summary.json
0001.mp3 Screenshot 2024-11-14 at 9.43.45 AM

0001.debug.json

Screenshot 2024-11-14 at 9.20.14 AM

navba-MSFT 25,075 Microsoft Employee

@Motyar D Welcome to Microsoft Q&A Forum, Thank you for posting your query here!

I tried to do a quick repro using the Batch Synthesis REST API with "wordBoundaryEnabled": true and it worked just fine. See below:

User's image

I am sharing my below request details, if that helps:

Request Headers:

PUT

https://westeurope.api.cognitive.microsoft.com/texttospeech/batchsyntheses/my-batch-synthesis-job-ssml-input-1?api-version=2024-04-01
  
Content-Type: application/json
Ocp-Apim-Subscription-Key: 62c2eXXXXXc65c6c

Request Body: .

{
  "inputKind": "SSML",
  "inputs": [
    {
      "content": "<speak version=\"1.0\" xml:lang=\"en-US\"><voice name=\"en-US-JennyNeural\">The rainbow has seven colors.</voice></speak>"
    }
  ],
  "properties": {
   "wordBoundaryEnabled": true,
    "destinationContainerUrl": "https://XXXXXXX.blob.core.windows.net/labaccel?sp=racwdli&st=2024-11-14T04:57:43Z&se=2024-11-14T12:57:43Z&spr=https&sv=2022-11-02&sr=c&sig=JL%2BjK5XZXXXXXXXuBMXI%3D",
    "decompressOutputFiles": true
  }
}

Action Plan:

I tested from the West Europe region. May I know which region is your Speech resource in ?
Could you please open the summary.json file, and confirm if you see below for do you see the below for wordBoundaryFileName property ?

{
  "jobID":  "b1bbf354-6ec2-4c16-9f57-81b6864fa798",
  "status":  "Succeeded",
  "results":  [
    {
      "texts":  [
        "<speak xmlns=\"http://www.w3.org/2001/10/synthesis\" ....</speak>"
      ],
      "status":  "Succeeded",
      "audioFileName":  "0001.wav",
      **"wordBoundaryFileName":  "0001.word.json",**
      "sentenceBoundaryFileName":  "0001.sentence.json",
      "properties":  {
        "audioSize":  "234044",
        "duration":  "PT4.8S",
        "durationInTicks":  "48750000"
      }
    }
  ]
}

I am using api-version, 2024-04-01. What is the api-version you are using ?
Could you please test with my above Request headers and request body and check if that helps ?
Could you please share your SSML file here so that I can also test from my end ?

. Awaiting your reply.

Motyar D 0 Reputation points

2024-11-14T08:18:36.8466667+00:00

$endpoint = "https://eastus.api.cognitive.microsoft.com/texttospeech/batchsyntheses";

$apiVersion = "2024-04-01";

Look at the file, there are no such keys
navba-MSFT 25,075 Reputation points Microsoft Employee

2024-11-14T08:21:47.27+00:00
@Motyar D Thanks for getting back.

May I know which region is your Speech resource in ?

Could you please share your SSML file here (in text format) so that I can also test from my end ?

Awaiting your reply.

Motyar D 0

Reason: eastus

SSML:

<speak version='1.0'

xmlns='http://www.w3.org/2001/10/synthesis'

xmlns:mstts='https://www.w3.org/2001/mstts' xml:lang='en-US'>

<voice name='en-US-Andrew2:DragonHDLatestNeural'>Hello there</voice>

<voice name='en-US-Emma2:DragonHDLatestNeural'>What? are you there? really?</voice>

<voice name='en-US-Brian:DragonHDLatestNeural'>Ok we are done!! bye haha</voice>

</speak>

1 answer

navba-MSFT 25,075 Reputation points Microsoft Employee

2024-11-14T10:00:25.45+00:00
@Motyar D Thanks for sharing the details. I was able to identify the cause of this issue.

.

Please note, not all voices support WordBoundary. The list is mentioned in this article.

.

Please update your SSML to include the voices which have support for wordBoundary. I have updated the SSML accordingly. Please use this SSML in your application and test this:

<speak version=\"1.0\" xmlns=\"http://www.w3.org/2001/10/synthesis\" xmlns:mstts=\"https://www.w3.org/2001/mstts\" xml:lang=\"en-US\"><voice name=\"en-US-AndrewMultilingualNeural\">Hello there</voice><voice name=\"en-US-EmmaMultilingualNeural\">What? Are you there? Really?</voice><voice name=\"en-US-BrianMultilingualNeural\">Ok, we are done! Bye, haha</voice></speak>

.

You should be able to see the 0001.word.json in your storage account.

**

Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Please sign in to rate this answer.
Motyar D 0 Reputation points

2024-11-14T10:05:16.42+00:00

Ok thanks, but how we can get the word boundary using the audio file as input? is that possible?

Which service and how we need to use?

Thanks

navba-MSFT 25,075 Reputation points Microsoft Employee

2024-11-14T10:37:11.1933333+00:00

@Motyar D Thanks for getting back. Yes, you can provide the audio file as an input for the below Speech REST API.

Use the fast transcription API with Azure AI Speech

Create a batch transcription

For the above answer, Please do not forget to "Accept the answer” and “up-vote” wherever the information provided helps you, this can be beneficial to other community members.
Sign in to comment

Use comments to ask for clarification, additional information, or improvements to the question.

Share via

[nnnn].word.json file not found in results wordBoundaryEnabled: true

1 answer

Your answer