Question 1

What does Gemini Audio Vision do?

Accepted Answer

It uses Google Gemini's native audio understanding to analyze a full audio file and returns a detailed text description covering spoken content, speaker changes, tone, background sounds, and music.

Question 2

What audio formats and lengths are supported?

Accepted Answer

Standard audio formats like MP3, WAV, and M4A are supported. Longer audio files take proportionally longer to process and cost more, since Gemini's token usage scales with audio duration.

Question 3

How is pricing calculated?

Accepted Answer

Pricing is token-based. Google bills audio input at a higher per-token rate than text, image, or video, so the per-run minimum is higher than Muapi's text-only Gemini endpoints. The actual cost is calculated from the API response and deducted from your wallet after each call.

Question 4

Does it transcribe speech verbatim?

Accepted Answer

Yes, if your prompt asks for a transcript. Gemini Audio Vision can produce verbatim transcripts, summaries, or structured analyses depending on how you phrase the prompt and system_prompt.

Question 5

Can I control the output format?

Accepted Answer

Yes. Use the system_prompt field to instruct the model to respond in a specific format, such as structured JSON, a bullet list, or a single sentence summary.

Question 6

How long does a request take?

Accepted Answer

Processing time depends on audio length and complexity. Short clips typically complete within seconds; longer audio may take up to a minute since the full file must be uploaded and ingested before analysis begins.