Should we remove Cloud TTS ? Since it's basically the same than google tts with vertex ADC auth

[Kamilbenkirane]

Open question
For cleaning purposes

[Seluj78]

I'd love more details so I can give an opinion :)

[Kamilbenkirane]

Decision: Keep Cloud TTS. Closing as won't-do.

Cloud TTS and Gemini TTS are different products serving different use cases:

• Cloud TTS (texttospeech.googleapis.com): 380+ voices, 75+ languages, MP3/WAV/OGG output, $4-30/1M chars. Traditional synthesis for IVR, accessibility, enterprise.
• Gemini TTS (generateContent): 30 voices, 24 languages, raw PCM only, LLM-powered prompt steering + multi-speaker. Creative/expressive content.

Only 8 voices overlap (Chirp HD). Neither is deprecated. Removing Cloud TTS would lose 370+ voices, 50+ languages, and encoded output formats.

What we should do instead (separate issues):

1. Add traditional Cloud TTS voice models (WaveNet, Neural2, Studio, Chirp HD)
2. Add Gemini TTS via generateContent path to unlock prompt control + multi-speaker