릴리스노트

Changelog

총건수

Abocado AI now has a dedicated Text-to-Speech (TTS) app, powered by Google's Gemini 3.1 Flash TTS. Type or paste text and turn it into natural, expressive speech — for video voiceovers, narration, and prototyping — without recording a thing.

What's new

  • A new TTS app joins the lineup, next to Image, Video, and Music.

  • Powered by Gemini 3.1 Flash TTS, Google's latest voice model tuned for controllability and quality.

What you can do

  • 30 prebuilt voices, 70+ languages — including regional accents (e.g., American and British English).

  • Steerable emotion & delivery — Gemini 3.1 Flash TTS supports 200+ audio tags like [whispers], [laughs], [excitement], [hope], plus pacing tags such as [slow], [fast], and [short pause] to make moments land.

  • Multi-speaker dialogue — generate back-and-forth conversations natively.

  • Plain-language direction — describe the tone ("calm narrator", "casual, conversational") and it adapts.

  • SynthID watermark — output is watermarked as AI-generated for transparency.

Who it's for

Anyone who needs a voice but not a microphone — short-form narration, product demos, accessibility, and script prototyping before a human voiceover.

How to use it

  1. Open Apps → Text-to-Speech (/ai/apps/tts).

  2. Enter the text you want spoken (add audio tags like [slow] or [excitement] to steer delivery).

  3. Choose a voice / language / style. «CONFIRM: Abocado TTS 앱에서 노출되는 정확한 컨트롤(보이스 선택·태그·멀티스피커)»

  4. Generate — your audio is ready to play and download.

Supported & limits

  • 30 voices · 70+ languages · SynthID‑watermarked output.

  • «CONFIRM: 글자 수 제한·출력 포맷·상업적 사용 라이선스»

FAQ

  • Can I control emotion or pacing? Yes — use audio tags like [excitement], [whispers], [slow], [long pause] in your text.

  • How many voices and languages? 30 prebuilt voices across 70+ languages with regional accents.

  • Is the audio watermarked? Yes, with SynthID.

Related

Google's Lyria 3 and Lyria 3 Pro are now available on Abocado AI. Unlike instrument-only generators, Lyria turns a text prompt into a full track with sung vocals and lyrics — so a written idea becomes a finished song. Both models are live in the music app, Popcone.

What's new

  • Lyria 3 — fast, high-fidelity text-to-music with vocals and lyrics, great for short tracks.

  • Lyria 3 Pro — the "thinking" model for longer, more structured songs and higher quality.

  • Both sit alongside the existing Popcone models, so you can pick the right engine per track.

What you can do

  • Sung vocals + lyrics, 8 languages — English, Korean, Japanese, German, Spanish, French, Hindi, Portuguese. Give it a theme and it writes and sings the lyrics, or supply your own.

  • Structured songs (Pro) — prompt for intros, verses, choruses, and bridges, and shape how the song progresses.

  • Studio-grade output — 44.1kHz stereo audio.

  • SynthID watermark — every track carries Google's invisible AI watermark for transparency.

Who it's for

Creators who write lyrics but don't have a vocalist — short-form video BGM, demos, and multi-language songs go from idea to sung track in one step.

How to use it

  1. Open Music → Popcone (/ai/apps/music).

  2. In the Model selector, choose Lyria 3 (fast) or Lyria 3 Pro (longer/structured).

  3. Enter a song title, then set Music Style, write a Prompt, or paste your Lyrics.

  4. Hit Generate Audio — your track appears in My Music.

Supported & limits

  • Vocal languages: 8 (listed above). Output: 44.1kHz stereo. Every track is SynthID‑watermarked.

  • «CONFIRM: Popcone에서 노출되는 세부 제어(구조/보컬 스타일)·최대 길이 표기·상업적 사용 라이선스»

FAQ

  • Can Lyria sing my lyrics? Yes — paste lyrics, or give a theme and let the model write and sing them.

  • Which languages can it sing? 8: English, Korean, Japanese, German, Spanish, French, Hindi, Portuguese.

  • Is the audio watermarked? Yes, with SynthID (Google's AI watermark).

Related