Why does TTS cognitive service sometimes return error 400 when using custom lexicon?

Naphat 25

When calling https://southeastasia.tts.speech.microsoft.com/cognitiveservices/v1? It returns the error "400 Synthesis failed. StatusCode: FailedPrecondition, Details: SSML parsing error: 0x8004801c - Wanted data exists in data file". However, the service sometimes works fine without any changes.

Lexicon File:

<lexicon xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="1.0" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="en-US"> <lexeme> <grapheme>THB</grapheme> <alias>Baht</alias> </lexeme> </lexicon>

Request Body:

<speak version="1.0" xml:lang="en-US" xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts"> <voice xml:lang="en-US" xml:gender="Female" name="en-US-AriaNeural"> <lexicon uri="https://pathtofile/en-US.xml" /> <mstts:silence type="Sentenceboundary" value="0ms" /> <mstts:silence type="Leading-exact" value="0ms" /> <mstts:silence type="Tailing-exact" value="0ms" /> <lang xml:lang="en-US">members enjoy a variety</lang> </voice> </speak>

Naphat 25 Reputation points

2024-09-14T17:40:57.8933333+00:00

as I keep testing it look like it's work fine on east.tts but always has error on southeastasia
Naphat 25 Reputation points

2024-09-14T17:48:58.28+00:00

interesting that it's happens only in southeastasia, but it work fine when I tried eastsasia
navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-16T06:40:38.08+00:00

@Naphat Apologies for the late reply. Welcome to Microsoft Q&A Forum, Thank you for posting your query here!

.

We are aware of this issue and the fix has been provided to this on the weekend. Please accept our sincere apologies for inconvenience caused due to this issue.

.

Could you please test again in the SouthEastAsia region and let me know if that works fine.

.

Awaiting your reply.
Sheng Zhao (SPEECH) 1 Reputation point Microsoft Employee

2024-09-16T13:06:11.48+00:00

which voices? can u share the ssml

Naphat 25

@navba-MSFT Yes, I'm still facing the same error: '400 Synthesis failed. StatusCode: FailedPrecondition, Details: SSML parsing error: 0x8004801c - Wanted data exists in data file.' This issue occurs in Southeast Asia and occasionally happens in East Asia.

Lexicon File:

<lexicon xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="1.0" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="en-US">
<lexeme>
<grapheme>THB</grapheme>
<alias>Baht</alias>
</lexeme>
</lexicon>

Request Body:

<speak version="1.0" xml:lang="en-US" xmlns="http://www.w3.org/2001/10/synthesis"
    xmlns:mstts="http://www.w3.org/2001/mstts">
    <voice xml:lang="en-US" xml:gender="Female" name="en-US-AriaNeural">
        <lexicon uri="https://pathtofile/en-US.xml" />
        <mstts:silence type="Sentenceboundary" value="0ms" />
        <mstts:silence type="Leading-exact" value="0ms" />
        <mstts:silence type="Tailing-exact" value="0ms" />
        <lang xml:lang="en-US">members enjoy a variety</lang>
    </voice>
</speak>

Naphat 25

this too

lexicon file:

<lexicon xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="1.0" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="zh-CN">
<lexeme>
<grapheme>THB</grapheme>
<alias>泰铢</alias>
</lexeme>
</lexicon>

request body:

<speak version="1.0" xml:lang="zh-CN" xmlns="http://www.w3.org/2001/10/synthesis"
    xmlns:mstts="http://www.w3.org/2001/mstts">
    <voice xml:lang="zh-CN" xml:gender="Female" name="zh-CN-XiaoxiaoNeural">
        <lexicon uri="https://
        <mstts:silence type="Sentenceboundary" value="0ms" />
        <mstts:silence type="Leading-exact" value="0ms" />
        <mstts:silence type="Tailing-exact" value="0ms" />
        <lang xml:lang="zh-CN">1. NARA 泰国料理，</lang>
    </voice>
</speak>

Naphat 25

sure,

Lexicon File:

<lexicon xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" version="1.0" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="en-US">
<lexeme>
<grapheme>THB</grapheme>
<alias>Baht</alias>
</lexeme>
</lexicon>

Request Body:

<speak version="1.0" xml:lang="en-US" xmlns="http://www.w3.org/2001/10/synthesis"
    xmlns:mstts="http://www.w3.org/2001/mstts">
    <voice xml:lang="en-US" xml:gender="Female" name="en-US-AriaNeural">
        <lexicon uri="https://pathtofile/en-US.xml" />
        <mstts:silence type="Sentenceboundary" value="0ms" />
        <mstts:silence type="Leading-exact" value="0ms" />
        <mstts:silence type="Tailing-exact" value="0ms" />
        <lang xml:lang="en-US">members enjoy a variety</lang>
    </voice>
</speak>

Naphat 25 Reputation points

2024-09-16T14:02:40.0833333+00:00

duplicated
navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-17T05:05:18.71+00:00

@Naphat Thanks for sharing the details. I am looking into this. I will get back once I have an update on this.
PeterS 0 Reputation points

2024-09-17T16:45:20.2233333+00:00

I have a similar error while generating TTS audio and avatar video with a custom lexicon at 'westeurope' region:

rpc error: code = InvalidArgument desc = Failed to synthesize audio, details: Connection was closed by the remote host. Error code: 1007. Error details: Status(StatusCode="FailedPrecondition", Detail="SSML parsing error: 0x8004801c - Wanted data exists USP state: TurnStarted. Received audio size: 0 bytes.

The error appears only for some voices but is not 100% reproducible. The affected voices include but are not limited to:

en-GB-SoniaNeural

en-GB-LibbyNeural

en-GB-RyanNeural

de-DE-AmalaNeural
DanG 0 Reputation points

2024-09-17T17:07:39.51+00:00

I am also getting this error when using a custom lexicon in the uksouth region. It mostly affects the voices en-GB-SoniaNeural and en-GB-RyanNeural, however other voices are sometime affected.

This is using REST API. Error is:
400 Synthesis failed. StatusCode: FailedPrecondition, Details: SSML parsing error: 0x8004801c - Wanted data exists in data file.

Request:
<?xml version="1.0" encoding="UTF-8"?><speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-GB"><voice name="en-GB-RyanNeural"><lexicon uri="https://public-url.com/my-lexicon.xml"/>some really simple text</voice></speak>

Lexicon:

<?xml version="1.0" encoding="UTF-8"?><lexicon version="1.0" xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="en-GB"><lexeme><grapheme>999</grapheme><alias>9 9 9</alias></lexeme></lexicon>

DanG 0

Also getting this error for uksouth with a variety of voices but mainly en-GB-SoniaNeural and en-GB-RyanNeural

Request is:

<?xml version="1.0" encoding="UTF-8"?><speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="en-GB"><voice name="en-GB-RyanNeural"><lexicon uri="https://public-url"/>some really simple text</voice></speak>

Lexicon is:

<?xml version="1.0" encoding="UTF-8"?><lexicon version="1.0" xmlns="http://www.w3.org/2005/01/pronunciation-lexicon" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" alphabet="ipa" xml:lang="en-GB"><lexeme><grapheme>999</grapheme><alias>9 9 9</alias></lexeme></lexicon>

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-18T04:51:49.2333333+00:00

@DanG @Peter Stafiychuk @Naphat We are aware of this issue. We are still investigating. All your above details with the SSML contents have been shared with the Product Owners. We are working on providing the fix. We will keep you posted.
Naphat 25 Reputation points

2024-09-18T12:20:02.3966667+00:00

Noted. Looking forward to further updates
navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-20T04:25:59.15+00:00

@Naphat @DanG @Peter Stafiychuk A quick follow-up to check if you had a chance to look at my above comment. Awaiting your reply.
DanG 0 Reputation points

2024-09-20T08:02:43.62+00:00

@navba-MSFT we have switched to westeurope. So far we have not encountered the issue when using our lexicon file.

Thank you for your assistance.
PeterS 0 Reputation points

2024-09-20T09:41:30.01+00:00

@navba-MSFT ,
Thanks for reaching out about the fixed issue!
I can confirm we don't have any reported cases today.

Thanks.
Naphat 25 Reputation points

2024-09-20T10:45:34.89+00:00

let me try it now
Naphat 25 Reputation points

2024-09-20T11:24:06.2533333+00:00

confirmed that I can't reproduce it in SEA anymore.
navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-20T12:18:18.98+00:00

@Naphat I am glad the issue is now resolved. For the below 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.

Accepted answer

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-19T08:46:43.83+00:00

@Naphat @DanG @Peter Stafiychuk Apologies for the late reply.

Please accept our sincere apologies for the inconvenience caused due to this issue.

.

Root Cause: There is a code bug which corrupt our engine when user sending refresh lexicon signal. All request with lexicon tag after it will failed to load lexicon files.

.

Fix update:

We have fixed the issue on SouthEastAsia, East US, East US 2 and West Europe. The fix is not yet deployed to all regions.

.

Could you please test this again in any of the above regions and let me know if you are still encountering this issue ?
Please sign in to rate this answer.

1 person found this answer helpful.
navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-20T12:18:43.78+00:00

@Naphat I am glad the issue is now resolved. 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.

Steven Verborgh 0 Reputation points

2024-09-24T10:01:35.4633333+00:00

Hi, it seems I'm still getting the errors as described above on the west europ region

Speech synthesizer returned with error: reason Error - errorCode: BadRequest - errorDetails: Connection was closed by the remote host. Error code: 1007. Error details: Status(StatusCode="FailedPrecondition", Detail="SSML parsing error: 0x8004801c - Wanted data exists USP state: TurnStarted. Received audio size: 0 bytes.. SSML Data: <speak xmlns="http://www.w3.org/2001/10/synthesis" xmlns:mstts="http://www.w3.org/2001/mstts" xmlns:emo="http://www.w3.org/2009/10/emotionml" version="1.0" xml:lang="de-DE"><voice name="de-DE-KatjaNeural"><lexicon uri="https://pitsassetsprdstorage.blob.core.windows.net/public/lexicon/lexicon_de_v001.004.000.xml"/><prosody rate="20%" pitch="0%">Gut gemacht, xxxx. Du hast schnell reagiert und Lily geht es besser.</prosody></voice></speak>

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-24T13:18:25.14+00:00

@Steven Verborgh This issue should be fixed in West Europe. Please test again and let me know.

Taylor Nelson 0 Reputation points

2024-09-26T06:57:33.71+00:00

It's been over a week, and this still isn't fixed in all regions, e.g. Australia East. When will this be fixed?

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-26T07:00:04.9966667+00:00

@Taylor Nelson Which region are you facing the issue now ? Also share your lexicon content.

Taylor Nelson 0 Reputation points

2024-09-26T07:03:38.1066667+00:00

@navba-MSFT as I said in my comment, Australia East. Here's my Lexicon.

<?xml version="1.0" encoding="utf-8"?> <lexicon xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://www.w3.org/2005/01/pronunciation-lexicon http://www.w3.org/TR/2007/CR-pronunciation-lexicon-20071212/pls.xsd" version="1.0" alphabet="ipa" xml:lang="en-AU" xmlns="http://www.w3.org/2005/01/pronunciation-lexicon"> <lexeme> <grapheme>AI's</grapheme> <phoneme>eɪ.aɪz</phoneme> </lexeme> <lexeme> <grapheme>ICT</grapheme> <phoneme>aɪ.siː.ˌtiː</phoneme> </lexeme> </lexicon>

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-26T07:05:40.9966667+00:00

@Taylor Nelson I will check and get back to you with an update. Apologies for the inconvenience caused.

navba-MSFT 24,985 Reputation points Microsoft Employee

2024-09-26T07:11:18.1766667+00:00

@Taylor Nelson The fix has just been deployed in Australia East. This issue should be fixed in next 40 minutes. Please try it after an hour.
Sign in to comment

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

Share via

Why does TTS cognitive service sometimes return error 400 when using custom lexicon?

0 additional answers

Your answer