Hi, i'm using fr-FR-RemyMultilingualNeural voice from azure. I need the agent to enumerate some strings that are customers ID's like AB123. My problem is the voice is switching to english while enumerating this, any way i can fix it ?

@hugopdvn Hello! It sounds like you're encountering a language switching issue with the Azure voice model when it comes to enumerating customer IDs. To maintain the French pronunciation for alphanumeric strings like "AB123," you can try using the Speech Synthesis Markup Language (SSML) to specify the language context for those strings. Here's an example of how you might format your SSML:

xml
<speak version="1.0" xmlns="http://www.w3.org/2001/10/synthesis" xml:lang="fr-FR">
    <voice name="fr-FR-RemyNeural">
        Voici l'ID du client: 
        <say-as interpret-as="characters">AB123</say-as>
    </voice>
</speak>

By using the

<say-as>

tag with the

interpret-as="characters"

attribute, you can instruct the voice model to spell out the ID character by character, which can help maintain the correct language pronunciation. If you're still experiencing issues, you might want to check the documentation for the specific voice model you're using or reach out to Azure support for more detailed guidance on handling multilingual content.

@Vapi Should I use this syntax into my script ?

@hugopdvn Yes, incorporating the SSML syntax into your script can help control the pronunciation and language behavior of the text-to-speech engine. By using SSML tags, you can specify that certain parts of your script should be read in French, even if they contain alphanumeric strings like customer IDs. Here's a quick guide on how to use the SSML syntax in your script: 1. Wrap your entire speech output in

<speak>

tags. 2. Specify the language and voice using the

<voice>

tag and the

xml:lang

attribute. 3. Use the

<say-as>

tag with the

interpret-as="characters"

attribute for alphanumeric strings to spell them out. Make sure to replace "AB123" with the actual customer ID you want to enumerate. If you have multiple IDs, you can repeat the

<say-as>

tag for each one. If you're not sure how to integrate SSML with your specific setup or if you need further assistance, please let me know, and I'll be happy to help!

that doesnt work

can you please share the call id along with actual vs expected response.

yes, for example here : bdb55646-c604-4384-ae16-74728166ef55 The bot starts to give me the ID in english

there is no english trasncription for this call id. { "role": "bot", "time": 1722519640868, "source": "", "endTime": 1722519644728, "message": "Bonjour, Audi By my car Lyon Nord à votre service,", "secondsFromStart": 1.04 }, { "role": "user", "time": 1722519651608, "endTime": 1722519652288, "message": "Oui alors", "duration": 1.1599998, "secondsFromStart": 11.78 }, { "role": "bot", "time": 1722519652948.001, "source": "", "endTime": 1722519665798, "message": "Je suppose donc que vous appelez pour connaître le statut de votre véhicule. Merci de votre appel. Votre demande concerne t l votre véhicule Peugeot quatre cent 6 avec la plaque d'immatriculation. Eux,", "secondsFromStart": 13.120001 }

@Shubham Bajaj the problem is the ID that looks like 'AD132FB' (2 letters followed by 3 digits, and then 2 letters again) is being prononced as an english (or german or whatever, it depends) word

i couldn't any word similar to this in transcription instead you can ask LLM to give numbers output in language you require.

allready done but that does not work 😦

can you please share the call id.

Marking as complete because of no response.