Speech to Text API — AI Transcription & Audio Analysis
Live2 modelsMuapi's speech-to-text API converts spoken audio into text through a single unified REST endpoint. Whisper handles accurate multilingual transcription with optional timestamped SRT/VTT subtitles, while Gemini Audio Vision goes further — reasoning about tone, speaker changes, and background sound instead of just transcribing words. Same submit-and-poll pattern as every other model on the platform, from $0.012 per minute.
OpenAI Whisper
Converts spoken audio into accurate written transcripts across dozens of languages. Optional segment-level timestamps or SRT/VTT subtitles for podcasts, meetings, and video captioning.
Gemini Audio Vision
Native audio understanding beyond plain transcription — analyzes speech, tone, speaker changes, and background sounds and returns a detailed text analysis for a given prompt.
What is a Speech to Text API?
A speech-to-text API converts spoken audio into written text via a single HTTP call — submit an audio file URL, receive a transcript back, no manual transcription or audio-engineering pipeline required.
Muapi exposes two audio-to-text endpoints through the same unified REST pattern used across every model on the platform — Whisper for accurate multilingual transcription, and Gemini Audio Vision for deeper audio understanding beyond plain text. One API key, one request/poll flow, pay-as-you-go pricing.
Key Capabilities
Multilingual Transcription
Whisper transcribes spoken audio across dozens of languages with clean, accurate output.
Timestamped Subtitles
Optional segment-level timestamps or ready-to-use SRT/VTT subtitle files for video captioning.
Native Audio Understanding
Gemini Audio Vision reasons about the full audio file end-to-end — speaker changes, tone and emotion, background sounds and music, not just spoken words.
Prompt-Driven Analysis
Ask Gemini Audio Vision a specific question about the audio and get a detailed text analysis back, rather than a flat transcript.
Meeting & Podcast Notes
Turn recorded meetings, podcasts, and interviews into searchable, structured text.
Voice-Driven Workflows
Feed transcripts into downstream text models, or pair with Muapi's text-to-speech models for a full voice-in, voice-out pipeline.
Speech to Text Model Comparison
| Model | Provider | Price | Best For |
|---|---|---|---|
| OpenAI Whisper | OpenAI | $0.012 / minute | Accurate multilingual transcription, subtitles |
| Gemini Audio Vision | $2.00/M in · $5.00/M out tokens | Deep audio understanding beyond transcription |
How to Transcribe Audio via API
- Pick a model — Whisper for a plain transcript, Gemini Audio Vision for a prompted analysis of the audio.
- Host your audio file — upload via
POST /api/v1/upload_fileor use any publicly reachable URL. - Submit the request —
POST /api/v1/{model-slug}with your audio URL and, for Gemini Audio Vision, a prompt describing what you want analyzed. - Poll for completion — check
GET /api/v1/predictions/{request_id}/resultuntilstatusiscompleted, then read the transcript or analysis.
Frequently Asked Questions
What is the Speech to Text API?
Muapi's speech-to-text API converts spoken audio into text via a single REST endpoint. Whisper covers accurate multilingual transcription; Gemini Audio Vision adds deeper audio understanding beyond plain transcription.
Can I get subtitles, not just a plain transcript?
Yes — Whisper supports optional segment-level timestamps or ready-to-use SRT/VTT subtitle files for video captioning.
How is Gemini Audio Vision different from transcription?
It reasons about the full audio end-to-end — speaker changes, tone, emotion, and background sounds — and returns a detailed text analysis for a prompt you supply, rather than a flat word-for-word transcript.
How much does speech-to-text cost?
Whisper is $0.012 per minute of audio (1-minute minimum). Gemini Audio Vision uses token-based billing at $2.00 per million input tokens and $5.00 per million output tokens.
Can I get Speech to Text API access right now?
Yes. Sign up at muapi.ai, create an API key from your dashboard, and start transcribing audio immediately — no waitlist required.