Which SSML tags are supported for Japanese TTS voices?

Question

Which SSML tags are supported for Japanese TTS voices?

Travis 20

Amazon Polly lists the tags it supports. Is there something similar for Azure TTS, or am I to assume all tags are supported by every language unless otherwise specified? This is important as we're building an SSML studio and need to restrict tags when certain voices are selected.

dupammi 8,615 Reputation points Microsoft External Staff

2024-08-08T08:18:35.5133333+00:00

Hi @Travis

Thank you for your question about SSML tag support for Japanese TTS voices in Azure.

Azure Text-to-Speech (TTS) supports a variety of SSML tags that can be used across different languages, including Japanese. However, not all tags or attributes may behave uniformly across languages or voices.

You can refer to the Azure Speech Service documentation for a detailed overview of supported SSML tags. This documentation outlines the various SSML tags that Azure TTS supports, such as <prosody>, <break>, <emphasis>, and more. It also includes details on how each tag's attributes, like rate, pitch, and volume, are handled.

It’s important to note that while many tags are generally supported, their behavior can differ based on the language and voice. For example, the <prosody> tag may interpret values differently depending on the specific voice used.

If your project requires precise control and validation of SSML tags for Japanese TTS, you might consider conducting tests with the specific tags and voices you intend to use to ensure compatibility.

I hope you understand. Thank you.
dupammi 8,615 Reputation points Microsoft External Staff

2024-08-09T05:23:32.4966667+00:00

Hi @Travis

We haven’t heard from you on the last response and was just checking back to see if you got a chance to check my above response. Thank you.
Travis 20 Reputation points

2024-08-09T05:35:43.51+00:00
Thank you for the quick response.
We're most interested in understanding the supported SSML tags for the Japanese voices:

Nanami

Keita

Aoi

Daichi

Mayu

Naoki

Shiori

If you have any details for these voices specifically, that would be greatly appreciated!
dupammi 8,615 Reputation points Microsoft External Staff

2024-08-13T06:28:10.19+00:00
Hi @Travis

Thank you for your follow-up query and Sorry for the delayed response.

To clarify the support for SSML tags in Microsoft Azure Text-to-Speech (TTS) for Japanese voices, it's important to reference the official Azure Speech Service documentation. Azure does support SSML for both neural and standard voices, but there are some differences in behavior across languages and voices.

SSML Tags Supported by Azure TTS

Azure TTS supports the following SSML tags across its voices, including Japanese:

<speak>: The root element of SSML, which contains the text to be spoken.

<voice>: Specifies the voice to be used.

<prosody>: Controls the pitch, rate, and volume of the speech.

<break>: Adds a pause between words or phrases.

<emphasis>: Emphasizes certain words or phrases.

<p>: Adds a pause between paragraphs.

<s>: Adds a pause between sentences.

<say-as>: Specifies how to interpret certain text (e.g., dates, times, or telephone numbers).

<phoneme>: Provides phonetic pronunciation of words.

<sub>: Substitutes a word or phrase with a different pronunciation.

Azure TTS and Japanese Voices

The following Japanese voices in Azure TTS are supported:

Nanami (Neural)

Keita (Neural)

Aoi (Neural)

Daichi (Neural)

Mayu (Neural)

Naoki (Neural)

Shiori (Neural)

Important Considerations

Behavior Variability: While these SSML tags are supported, their effects might vary slightly depending on the language and specific voice. For example, how <prosody> affects pitch and rate can differ from one voice to another.

Testing: It's recommended to test SSML tags with each voice to ensure the desired output, particularly for Japanese TTS.

For specific behavior or limitations of certain voices or languages, you should test the voices directly using the Azure TTS API.

I hope this helps. Thank you.

Your answer

dupammi 8,615 Reputation points Microsoft External Staff

2024-08-08T08:18:35.5133333+00:00

Hi @Travis

Thank you for your question about SSML tag support for Japanese TTS voices in Azure.

Azure Text-to-Speech (TTS) supports a variety of SSML tags that can be used across different languages, including Japanese. However, not all tags or attributes may behave uniformly across languages or voices.

You can refer to the Azure Speech Service documentation for a detailed overview of supported SSML tags. This documentation outlines the various SSML tags that Azure TTS supports, such as <prosody>, <break>, <emphasis>, and more. It also includes details on how each tag's attributes, like rate, pitch, and volume, are handled.

It’s important to note that while many tags are generally supported, their behavior can differ based on the language and voice. For example, the <prosody> tag may interpret values differently depending on the specific voice used.

If your project requires precise control and validation of SSML tags for Japanese TTS, you might consider conducting tests with the specific tags and voices you intend to use to ensure compatibility.

I hope you understand. Thank you.
dupammi 8,615 Reputation points Microsoft External Staff

2024-08-09T05:23:32.4966667+00:00

Hi @Travis

We haven’t heard from you on the last response and was just checking back to see if you got a chance to check my above response. Thank you.
Travis 20 Reputation points

2024-08-09T05:35:43.51+00:00

Thank you for the quick response.
We're most interested in understanding the supported SSML tags for the Japanese voices:

Nanami

Keita

Aoi

Daichi

Mayu

Naoki

Shiori

If you have any details for these voices specifically, that would be greatly appreciated!
dupammi 8,615 Reputation points Microsoft External Staff

2024-08-13T06:28:10.19+00:00

Hi @Travis

Thank you for your follow-up query and Sorry for the delayed response.

To clarify the support for SSML tags in Microsoft Azure Text-to-Speech (TTS) for Japanese voices, it's important to reference the official Azure Speech Service documentation. Azure does support SSML for both neural and standard voices, but there are some differences in behavior across languages and voices.

SSML Tags Supported by Azure TTS

Azure TTS supports the following SSML tags across its voices, including Japanese:

<speak>: The root element of SSML, which contains the text to be spoken.

<voice>: Specifies the voice to be used.

<prosody>: Controls the pitch, rate, and volume of the speech.

<break>: Adds a pause between words or phrases.

<emphasis>: Emphasizes certain words or phrases.

<p>: Adds a pause between paragraphs.

<s>: Adds a pause between sentences.

<say-as>: Specifies how to interpret certain text (e.g., dates, times, or telephone numbers).

<phoneme>: Provides phonetic pronunciation of words.

<sub>: Substitutes a word or phrase with a different pronunciation.

Azure TTS and Japanese Voices

The following Japanese voices in Azure TTS are supported:

Nanami (Neural)

Keita (Neural)

Aoi (Neural)

Daichi (Neural)

Mayu (Neural)

Naoki (Neural)

Shiori (Neural)

Important Considerations

Behavior Variability: While these SSML tags are supported, their effects might vary slightly depending on the language and specific voice. For example, how <prosody> affects pitch and rate can differ from one voice to another.

Testing: It's recommended to test SSML tags with each voice to ensure the desired output, particularly for Japanese TTS.

For specific behavior or limitations of certain voices or languages, you should test the voices directly using the Azure TTS API.

I hope this helps. Thank you.

Share via

Which SSML tags are supported for Japanese TTS voices?

Your answer