Gemini 3.8 Flash-Lite TTS

Gemini 3.8 Flash-Lite TTS is Google’s fast, cost-efficient text-to-speech model. It ranks second on Hume AI’s Overall Quality Index, right behind google/gemini-3.8-flash-tts, at a lower price. It supports 101 languages.

Use it for voice agents, customer support bots, read-aloud features, dubbing, and other high-volume work. Google recommends it as the drop-in upgrade from google/gemini-3.1-flash-tts.

It takes exactly the same inputs as Gemini 3.8 Flash TTS. Pick Flash TTS instead when you need studio-grade acting, dialect work, two-speaker scenes with lots of backchanneling, or long-form narration.

How it works: cast, direct, speak

• Cast (voice): who is speaking. Pick one of 30 studio voices.
• Direct (style): how the line is delivered, like calm customer support agent or whispered urgently.
• Speak (text): exactly what they say. Everything in text gets spoken, except inline tags in <angle brackets> and backchannels in |pipes|.

Inputs

Start with no style

Try your script with an empty style first. Most text needs no direction, because the model reads punctuation, context, and the voice’s own character. Add a few words of style only when a line needs a nudge.

text:  Thanks for calling! <short pause> I can help you reset your password in just a minute.
voice: Achird
style: calm, friendly customer support agent

For voice agents, make one call per turn, leave style empty or use the same short style for the whole conversation, and let the voice carry the character.

Vocal bursts and pauses

Tags in angle brackets make a sound once, at the exact spot you put them:

• Breathing: <breath>, <heavy breath>, <exhales>, <pant>, <sigh>
• Laughter: <laugh>, <chuckle>, <giggle>, <snicker>, <cackle>
• Reactions: <gasp>, <cough>, <sneeze>, <throat-clearing>, <snort>
• Sounds: <sob>, <cry>, <whimper>, <groan>, <yawn>
• Pauses: <short pause>, <long pause>, <tsk>, <argh>

CAPITALS add stress to a word, and ellipses (...) add hesitation. If your text isn’t in English, keep the tags in English anyway.

Custom pronunciation

Write tricky names in the International Phonetic Alphabet between slashes:

text: Let me introduce you to my colleagues /niːv/ and /ʃɪˈvɔːn/.

Two-speaker dialogue

Set second_voice and write each turn on its own line as Name: what they say, using exactly two names. The first name gets voice and the second gets second_voice. To direct one line, write Name (style): what they say. Wrap a short reaction in pipes, like |seriously?|, and the other speaker says it in the background.

voice:        Puck
second_voice: Kore
text: |
  Host: So you're telling me the whole app runs on a single server? |seriously?|
  Guest: <laugh> One server, three cron jobs, and a lot of optimism.
  Host (laughing): That's the most honest answer we've ever had on this show.

Tips

• Keep style short. A few words beat a paragraph. Long direction blocks make the voice drift.
• Don’t use style to change who’s speaking. Age, gender, and accent come from the voice, so pick a different voice instead.
• Write how people talk. Fillers and hesitations (“oh, uh, yeah… hm”) sound more natural.
• Split long responses into shorter turns instead of reaching for stronger style prompts.

Voices

Languages

The model detects the language from your text. It supports 101 languages. A few that Flash TTS supports are missing here, including Finnish, Swedish, Thai, Lithuanian, and Slovenian. See Google’s language list.

Upgrading from Gemini 3.1 Flash TTS

• Move direction out of the text and into style. The prompt input is gone.
• Tags now use angle brackets (<laugh>), not square brackets ([laughing]).
• language_code is gone, because the model detects the language itself.

Safety

Every clip includes an inaudible SynthID watermark that marks it as AI-generated.

Learn more

• Gemini 3.8 Flash TTS developer guide
• Google’s speech generation docs