Gemini 2.5 Flash TTS

What it does

Gemini 2.5 Flash TTS is a text-to-speech model from the Gemini provider. You give it written text and it returns spoken audio. On Pipe2 it runs inside the Audio Generator pipeline, where you pick from dozens of voices across 70+ languages and can attach style directions that steer tone, accent, and pacing. It is positioned as the faster, cheaper synthesis option, aimed at drafts and short clips rather than long final renders.

When to use it

• Running Audio Generator to turn a written script into narration or voiceover without recording anything yourself.
• Iterating on wording: because it is the faster, cheaper of the synthesis options, it suits repeated takes while you are still editing the script.
• Producing short clips (stingers, line reads, prompts) where you want the result back quickly.
• Testing voice and language choices across the 70+ supported languages before committing to a longer render.
• Trying style directions for tone, accent, and pacing to hear how a delivery instruction changes the read.

What to know before you run it

Billing is metered on generated output tokens (the wholesale unit is quoted per roughly 1500 output tokens), so cost scales with how much speech you generate. A long script costs more than a short one, and the practical way to control spend is to shorten the text or split it into shorter segments and generate them separately. The input is text, plus the voice and style direction you choose in Audio Generator. The trade-off for the faster, cheaper tier is that it targets drafts and short clips. One worked example is published on Pipe2 if you want to see the inputs it was run with.

Also known as

You may see this model referred to as Gemini 2.5 Flash Text-to-Speech, Gemini 2.5 Flash Preview TTS, or gemini-2.5-flash-preview-tts; the shorthand "Gemini-TTS" also refers to it.