How do I generate speech with SSML tags in an API?

To generate speech with SSML in an API, you generally:

1. Send plain text or SSML-formatted XML to a text-to-speech endpoint.
2. Set the request to indicate SSML, not plain text.
3. The TTS engine parses the SSML tags and produces speech accordingly.

Typical flow

1) Write SSML

Example:

<speak>
  Hello <break time="500ms"/> world.
  <prosody rate="slow" pitch="+2st">
    This part is slower and higher pitched.
  </prosody>
</speak>

2) Call the TTS API

Most APIs expose a field like:

• input.text
• input.ssml
• text_type: "ssml"
• input_format: "ssml"

Example JSON request

{
  "voice": "en-US-JennyNeural",
  "input": {
    "ssml": "<speak>Hello <break time='500ms'/> world.</speak>"
  },
  "output_format": "audio/mp3"
}

Or sometimes:

{
  "text": "<speak>Hello <break time='500ms'/> world.</speak>",
  "textType": "ssml"
}

3) Receive audio

The API usually returns:

• a binary audio stream, or
• base64-encoded audio, or
• a URL to download the generated file

Common SSML tags

Useful tags include:

• <break time="300ms"/> — pause
• <prosody rate="slow" pitch="+2st">...</prosody> — adjust speech
• <emphasis level="strong">...</emphasis> — emphasize words
• <say-as interpret-as="characters">API</say-as> — spell out text
• <voice name="...">...</voice> — choose a voice
• <lang xml:lang="es-ES">...</lang> — switch language

Example

<speak>
  <voice name="en-US-JennyNeural">
    Hello there. <break time="300ms"/>
    Welcome to the demo.
  </voice>
</speak>

Things to watch for

• SSML must be well-formed XML.
• Some APIs require the root <speak> tag.
• Not all SSML tags are supported by every vendor.
• You may need to escape special characters like &, <, and > inside text.

If you want, I can show you how to do this for a specific provider like AWS Polly, Google Cloud TTS, Azure Speech, or OpenAI.