Azure-docs: Is there any way to validate my data before importing it?

[rajatvijay]

2 times my import failed and the UI doesn't show any reason.

---

Document Details

⚠ Do not edit this section. It is required for docs.microsoft.com ➟ GitHub issue linking.

• ID: 50d8bb98-fbcd-7562-99ed-54211fd9afd3
• Version Independent ID: bd23e619-aee5-2b6e-b2d7-03ab5f7302af
• Content: Tutorial: Create an acoustic model with the Speech Service
• Content Source: articles/cognitive-services/Speech-Service/how-to-customize-acoustic-models.md
• Service: cognitive-services
• GitHub Login: @PanosPeriorellis
• Microsoft Alias: panosper

[AdamS-MSFT]

@rajatvijay Thank you for the valuable feedback,we are investigating the issue.

[YutongTie-MSFT]

@erhopf Hi, could you please share the direction about who I should reach out for product issue/feedback of Speech Services? Thank you.

[erhopf]

@YutongTie-MSFT - I've reached out to the engineering team and will let you know as soon as I have an answer.

[rajatvijay]

So what I have figured is:

1. If the data is not in the right format as described in the docs, the process fails. But it doesn't explicitly tells what the problem exactly is.
2. If the data is in the correct format but some data points were not being processed, maybe because of some error specifically in them, then it explicitly tells (on the UI) the exact problem.

[erhopf]

@YutongTie-MSFT can you please assign this to @LeonRomaniuk.

@LeonRomaniuk - Who should we reach out to investigate improving error handling?

[YutongTie-MSFT]

@erhopf Thanks Erik.

[LeonRomaniuk]

You should reach out to Wolfgang Manousek.

[YutongTie-MSFT]

@LeonRomaniuk Got it, thanks.

@wolfma61 Hi Wolfgang, could you please take a look of this issue? Thanks.

[ascortez]

Any docs on the Accuracy test? Would love to see feedback on how to improve if we see a poor error rate. Because we do not know how the tests are being done, improving the rate becomes a game of guessing and checking.

[wolfma61]

sorry - not the expert on custom speech models.
@PanosPeriorellis can probably point to the right people

[emazzotta]

It would be VERY helpful if any sort of failure information could be shown.

I tried to boil down the error by uploading a single zipped .wav-file with the following properties:

RIFF (little-endian) data, WAVE audio, Microsoft PCM, 16 bit, mono 8000 Hz

Accordingly, I uploaded a transcript.txt file with the following line:

gf1_01_001.wav  contact geneva one two eight decimal one five good bye

Even with this very simple use case I just get the status "Failed" and clicking on "Details" does not reveal anything helpful.

[ascortez]

Experienced same issue. I believe that my audio formatting is correct because I was able to use the same audio files for testing on the Endpoint tab later. Which leads me to believe it was a transcription file error. But other that using tabs to space, and all lower case, not sure what the issue is.

This service is nearly unusable due to this issue.

[PanosPeriorellis]

The data is validated at import. We do not provide another a separate tool for data validation.

From: Ash Cortez [e-mail]
Sent: 26 February 2019 11:09
To: MicrosoftDocs/azure-docs [e-mail]
Cc: Panos Periorellis [e-mail]; Mention [e-mail]
Subject: Re: [MicrosoftDocs/azure-docs] Is there any way to validate my data before importing it? (#24283)

Experienced same issue. I believe that my audio formatting is correct because I was able to use the same audio files for testing on the Endpoint tab later. Which leads me to believe it was a transcription file error. But other that using tabs to space, and all lower case, not sure what the issue is.

This service is nearly unusable due to this issue.

—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHubhttps://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FMicrosoftDocs%2Fazure-docs%2Fissues%2F24283%23issuecomment-467570962&data=02%7C01%7Cpanos.periorellis%40microsoft.com%7Cf33d5b27341a41472fa308d69c1dec42%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636868049662553578&sdata=EpBKOLBsdCjL6lQv3noj14N5yTpbwmLZ4eIlbFpTWQM%3D&reserved=0, or mute the threadhttps://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fnotifications%2Funsubscribe-auth%2FAYK9kha7uPDgUK5NMUSg65xMchFZh3RBks5vRYZkgaJpZM4ajQQP&data=02%7C01%7Cpanos.periorellis%40microsoft.com%7Cf33d5b27341a41472fa308d69c1dec42%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C636868049662553578&sdata=hVVwXgv3u08R0aFh8848cEswxjKnlcdgKJJRjlbAJ%2Bs%3D&reserved=0.

[tchristiani]

please-close