fix: pass speed and emotion params through tts() and tts_to_file()

[ivanVgusev]

Problem

speed passed to tts_to_file() and
tts() was silently consumed and never
forwarded to the underlying model. As a
result:

  tts.tts_to_file(..., speed=1.5)

had no effect on XTTS v2, even though
Xtts.inference() fully supports
speed.

Root cause

• tts_to_file() did not pass
  speed/emotion to self.tts()
• tts() did not pass speed to
  self.synthesizer.tts()

synthesizer.tts() forwards **kwargs →
Xtts.synthesize() → inference(),
so the fix only required two missing
arguments in the API layer.

Changes

• tts(): add speed=speed to
  synthesizer.tts() call
• tts_to_file(): add speed=speed, emotion=emotion to self.tts() call
• Update docstrings: speed works with
  XTTS v2, not only Coqui Studio models

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

ivanVgusev seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}

[eginhard]

You can use our fork (available via pip install coqui-tts) where this is fixed since v0.26.0. This repo is not maintained anymore.