Azure TTS Plays Wrong Device

This behavior usually comes down to how Audio.sayEZB() routes audio versus how the Azure Text To Speech skill is configured.

What’s happening

• Audio.sayEZB(...) is the built-in ARC audio command for the EZ-B speaker path.
• The Azure Text To Speech skill only takes over Audio.say() / Audio.sayEZB() if its replacement option is actually being used by that exact skill instance.
• Since your soundboard is explicitly set to IoTiny #1 and works, that tells us the speaker path itself is fine.
• But if Azure TTS is still playing on EZB #0, then one of these is likely true:
  1. The Azure TTS skill is not actually the instance replacing the audio commands you’re calling.
  2. The ezbIndex is being ignored because the replacement skill is sending audio to its configured output device.
  3. The skill is configured to output to a different device than you expect.

Important detail

When "Replace Audio Script Commands" is enabled, the Azure TTS skill can override Audio.say() / Audio.sayEZB(). In that case, the output follows the skill’s own configuration, not necessarily the ezbIndex you pass into Audio.sayEZB().

So the 1 in this:

Audio.sayEZB("test de voix francaise", 1);

may not be controlling the Azure TTS output the way you expect once replacement is enabled.

What to check

1) Open the Azure Text To Speech skill settings

Click the kebab / menu icon on the Azure Text To Speech skill and verify:

• Speak out of EZB is enabled if you want robot audio
• The skill is not set to output to the PC if you want IoTiny audio
• The skill instance is the one you intend to use

2) Test the skill directly

Try using the Azure skill’s own ControlCommand instead of Audio.sayEZB():

controlCommand("Azure Text To Speech", "speak", "test de voix francaise");

Or, if you want to wait until it finishes:

controlCommand("Azure Text To Speech", "speakWait", "test de voix francaise");

If that still outputs to EZB #0, then the Azure skill configuration is likely pointing there.

3) Verify the skill name

If you renamed the Azure skill, the controlCommand() target name must match exactly.

4) Check whether multiple audio/TTS skills are active

If more than one speech-related skill is present, one may be intercepting the command path.

Best next test

Try this sequence:

controlCommand("Azure Text To Speech", "setVoice", "fr-FR-JacquelineNeural");
sleep(500);
controlCommand("Azure Text To Speech", "speakWait", "test de voix francaise");

If that works but still goes to the wrong EZ-B, then the output device is controlled in the Azure TTS skill config, not by Audio.sayEZB().

My guess

Because Soundboard v4 is explicitly tied to IoTiny #1 and works, but Azure TTS replacement is still going to EZB #0, I suspect the Azure skill is either:

• configured to output to EZB #0, or
• overriding the Audio.sayEZB() call and using its own device routing.

A couple questions

1. In the Azure Text To Speech skill settings, is "Speak out of EZB" checked?
2. Is the Azure skill named exactly "Azure Text To Speech" in the project?
3. Do you have the Azure skill added only once, or is there more than one TTS/speech skill in the project?

If you want, I can help you build the exact ARC script and skill setup so the French voice always comes out of IoTiny #1. Also, ARC’s built-in AI Assistant can often rewrite this kind of script for you right inside the editor:
https://synthiam.com/Products/AI-Assistant

controlCommand("Azure Text To Speech", "setVoice", "fr-FR-JacquelineNeural");
sleep(500);
controlCommand("Azure Text To Speech", "speakWait", "test de voix francaise");

it goes to the pc. the "speak out of ezb " , same behavior if it is checked or not. only one azure skill. soundboard v4 is loaded as well. looks like, the output is ok on ezb #0 or pc but not on another ezb #1

good news, i removed the skill azure and I loaded the skill speech settings, choose the right voice , check enabled and it works. thanks for your support