Top 3 new model launches at Gemini Audio at Night
Read the Portuguese translation →
Author: Google for Developers · Watch on YouTube
Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.
Original language: English
Original video: Top 3 new model launches at Gemini Audio at Night · Channel: Google for Developers · 01:06 · Watch on YouTube
Topics: Software engineering with AI, Multimodal models
Summary
Section titled “Summary”Google DeepMind has released new audio models and features for developers, including personalized text-to-speech, live translation, and expanded Gemini Live capabilities. The text-to-speech models let developers shape a voice’s emotion, resonance, and pauses. Live translation can convert video or audio input into any of 100 languages. Gemini Live adds multimodal understanding, live interactions, and function and tool calls for developer projects.
Key points
Section titled “Key points”- Google DeepMind has released new audio models for live translation, live interactions, and voice customization.
- The text-to-speech models support voice personalization.
- Developers can guide a voice’s emotions, resonance, and pauses.
- Live translation accepts video or audio input.
- Live translation supports 100 different languages.
- Gemini Live models offer multimodal understanding and live interactions.
- Function calls and tool calls can be incorporated into developer projects.
- The video invites developers to build with the new audio models.
Chapters
Section titled “Chapters”| Start | Chapter | Summary |
|---|---|---|
| 00:01 | New audio models and voice personalization | The video introduces Google DeepMind’s new audio models and explains how text-to-speech voices can be customized with emotions, resonance, and pauses. |
| 00:34 | Live translation | Live translation can take video or audio input and translate it into any of 100 languages. |
| 01:06 | Building with the new audio models | The video closes by encouraging developers to build with the new audio models. |
Quotes
Section titled “Quotes”Here are the top three things that you should check out. — 00:01
you can create a voice and guide it to have specific emotions, specific resonance, and even to incorporate things like pauses. — 00:01
change it to any of a hundred different languages that you can imagine. — 00:34
bake them into all of your developer projects. — 00:34
We can’t wait to see what you build — 01:06
Concepts and entities
Section titled “Concepts and entities”Google DeepMind, text-to-speech models, live translation, Gemini Live
Full transcript (original language)
Section titled “Full transcript (original language)”00:01 Google DeepMind just released some amazing new audio models everything from live translation to live interactions to things like voice customization and personalization. Here are the top three things that you should check out. Our new text-to-speech models are capable of doing really compelling voice personalization, which means that you can create a voice and guide it to have specific emotions, specific resonance, and even to incorporate things like pauses. - Tonight, fog rolling in from the Pacific. - We’ve also just recently released something called
00:34 live translation, which allows you to have a video input feed or an audio input feed and change it to any of a hundred different languages that you can imagine. - The salmon avocado, please. - Eeto, supaishii tsuna to saamon abokado o onegaishimasu. - Yeah! - We also have new features added into our Gemini Live models. So you can take the multimodal understanding, the live interactions, things like function calls and tool calls, and bake them into all of your developer projects. - That full chest turn is the main thing to keep practicing.
01:06 Nice work today. - We can’t wait to see what you build, especially what you build with these new audio models.