Gemini 2.5 Flash Native Audio Preview 12-2025
Gemini 2.5 Flash Native Audio: instant audio generation from text.
A tradução deste card ainda não está disponível; parte do conteúdo aparece em inglês.
- Janela de contexto (tokens)
- 131.072
Gemini 2.5 Flash Native Audio Preview 12-2025 is a fast, efficient model designed for generating audio from text prompts. It excels at creating a wide variety of sounds, from realistic environmental audio to stylized sound effects and simple melodies, all with low latency.
Use this model when you need quick audio prototypes, background sounds for videos, or creative sound design ideas. Its speed makes it ideal for iterative workflows where you want to experiment with different audio concepts rapidly.
What sets it apart is its native audio generation capability, meaning it produces audio directly without needing separate text-to-speech or music generation pipelines. This streamlined approach allows for fast, coherent results that match your prompt closely.
- Sound effects for games and videos
- Background ambience for storytelling
- Quick audio prototypes for creative projects
- Musical loops and simple melodies
- Educational audio examples and narration
Produces audio clips in seconds, enabling rapid iteration.
Handles sound effects, ambience, and simple music from one model.
Closely matches the described audio style and content.
- Be specific with sounds, like 'gentle rain on a window' instead of just 'rain'.
- Use descriptive adjectives for mood, e.g., 'cheerful piano melody' or 'ominous wind'.
- Keep prompts concise for best results; avoid overly long descriptions.
- Try multiple variations of a prompt to explore different audio outputs.
- Limited to short audio clips, not full songs or long narratives.
- May struggle with complex musical arrangements or multiple simultaneous instruments.
- Generated audio quality can vary for very abstract or unusual prompts.
Gentle birdsong, rustling leaves, and a soft breeze in a magical forest at sunrise
Classic 8-bit video game coin collect sound effect, bright and cheerful
Soothing sound of gentle waves lapping on a sandy shore with seagulls in the distance
A number that makes results reproducible: same seed and prompt give the same audio.
Nucleus sampling: limits the model to consider only the most likely next sounds, balancing creativity and coherence.
Sets the length of the generated audio clip in seconds.
Controls randomness: lower values make output more predictable, higher values add variety.
How closely the output follows your prompt: higher values mean stricter adherence.
Os preços exibidos são o que você paga na Muitto. Sem surpresas.
What kind of audio can I generate?
You can generate sound effects, ambient sounds, and simple musical clips.
How long does it take to generate audio?
Generation is typically very fast, often taking just a few seconds.
Can I generate music with lyrics?
No, this model is designed for instrumental audio and sound effects, not singing.