gemini-3.8-flash-tts — Speech AI Model
Gemini turns written text into a spoken voiceover and hands back a WAV clip — pick a prebuilt voice and it detects Korean, English and more from the text itself. Cloud-side, so no GPU; bring a Google AI API key.
What it does
Gemini TTS reads text aloud and returns a WAV clip on the sound port. A language model does the speaking rather than a phoneme engine, so tone and pacing follow from the sentence instead of coming out flat.
Six prebuilt narrator voices, and they are language-agnostic - the spoken language follows the text. Left on Auto the language is inferred; set Korean or English when the text is mixed enough to be ambiguous. Nothing runs locally, and it needs a Google AI key.
When to use it
- A cold-open voiceover line to put in front of a short-form clip
- Narrating a script that an earlier node in the flow just wrote
- Korean and English narration without changing voice or node
- Long-form narration where the point is to avoid local GPU time
Details
- CategorySpeech
- Year2026
- LicenseGoogle Cloud Terms
Compliance & Provenance
- ProviderGoogle · GPAI
- EU AI Act RiskLimited Risk
- Art. 50 TransparencyRequired — AI-generated outputs are marked
Inputs & Outputs
- TextInputstring
Text to speak (the narration script)
- SoundOutputsound
Synthesized voiceover audio (WAV)
Parameters
- modelstringdefault Gemini 3.8 Flash TTSoptions: Gemini 3.8 Flash TTS, Gemini 3.8 Flash-Lite TTS, Flash TTS, +2 more
Model variant
Which Gemini TTS model to call. 'Gemini 3.8 Flash TTS' (default) — studio-grade voice and expressive delivery. 'Gemini 3.8 Flash-Lite TTS' — cheaper and lower latency. 'Flash TTS', 'Pro TTS' and 'Gemini 3.1 Flash TTS' are kept for saved flows and now run Gemini 3.8 Flash TTS (Google shuts the older previews down on 2026-11-17).
- voicestringdefault Koreoptions: Kore, Puck, Charon, Aoede, Leda, Fenrir
Prebuilt narrator voice.
Prebuilt Gemini voice used for synthesis. Voices are language-agnostic (the spoken language follows the input text). Kore (default) is a balanced narrator.
- languagestringdefault Autooptions: Auto, Korean, Englishadvanced
Spoken language hint (Auto follows the text).
Optional BCP-47 language hint (e.g. ko-KR, en-US). 'Auto' (default) lets the model infer the language from the input text.
- speedfloatdefault 1range 0.5–2
Speaking speed. 1.0 = as generated; >1 faster, <1 slower. Pitch is kept.
Playback speed of the synthesized voice, 0.5-2.0 (default 1.0). The audio is re-timed after synthesis with its pitch kept, so 1.1 means exactly 10% faster. Values outside the range are clamped.
Limitations
- One WAV clip per run. There is no pitch or per-word timing control on this node; speed is applied after synthesis by re-timing the clip.
- The language setting offers Auto, Korean and English; anything else has to arrive through Auto.
- 'Flash TTS', 'Pro TTS' and 'Gemini 3.1 Flash TTS' now all run Gemini 3.8 Flash TTS - Google retires the older models on 2026-11-17.
- Runs on your own Google AI API key. Without one configured in the workspace the node cannot run.
Tags
Alternatives in Audio Models
- Music (Lyria)
Google Lyria: text (and optional image) to music — full songs with vocals and lyrics, or 30-second clips. Needs API key.
- Text to Music (ACE-Step 1.5)
ACE-Step 1.5: generate music from a style prompt, with optional lyrics, key, tempo and time signature. 48 kHz stereo, 5-600s.
- Text to Sound Effect (MOSS-SoundEffect v2)
MOSS-SoundEffect v2.0: generate a sound effect from an English description. 48 kHz mono, 0.5-30s.
- Text to Speech (Qwen3 Voice Design)
Qwen3-TTS VoiceDesign: on-device text-to-speech with a voice you describe in words. No API key.
- Text to Speech (Qwen3)
Qwen3-TTS: on-device text-to-speech voiceover (Korean/English). No API key.