AI Studio speech playground does not play generated audio. Short text sometimes speaks. A normal Vietnamese paragraph only plays the first syllable (“Tôi”) for under a second, then stops. The player stays at 0:00. No error is shown in the UI.
Model: gemini-3.8-flash-tts
Page: https://aistudio.google.com/generate-speech
Project: gen-lang-client-0584548..
Voice: Riko (also failed with Charon)
Account: API key linked (USING KEY)
Network:
- POST MakerSuiteService/GenerateContent
- URL host: alkalimakersuite-pa.clients6.google.com
- Status: 200 OK
- Response contains audio/l16; rate=24000 and a media id starting with v1_Chdj…
- The audio payload visible in DevTools is only “AAAAAAAAAAAAAAAA” (silence / a few samples)
- A follow-up media request is named “null”, size 0.0 kB, status 200
- Because the player loads “null”, playback dies immediately
This worked on the previous free playground: I generated full 10–12 minute narrations (Vietnamese and other languages) with Charon. The new 3.8 playground no longer has the style-instructions box, and long text no longer returns playable audio.
Please check the speech playground playback path for gemini-3.8-flash-tts. I can provide screenshots of the GenerateContent Headers (200) and Response (audio/l16 plus null media request). I will not post the API key.
