Zoom AI Services: Scribe API

The World’s Most Accurate Speech-to-Text API

Quality transcription you can trust, powered by Zoom’s ASR Model Pro.

Ranked as a top performing model on the HuggingFace Open ASR Leaderboard, Scribe API offers both Fast Sync and Batch Transcription services, built on the same ASR powering millions of Zoom meetings every day.

See Docs Get API Key

Capabilities

Everything You Need for Speech-to-Text

From near real-time to batch processing, Zoom Scribe API delivers transcription services with high accuracy and speed.

Fast Sync Transcription

Access synchronous, low-latency transcription for individual audio files. Process one file at a time with immediate response after completion.

Batch Transcription

Process pre-recorded audio and video files at scale. Support for MP3, WAV, MP4, FLAC, OGG, and more with automatic format detection.

Word-Level Timestamps

Get precise start and end times for every word, enabling perfect subtitle generation, audio search, and content indexing.

Use Your Own Storage

Store your transcripts securely in your own AWS S3 bucket for enhanced data control and compliance.

Speaker Diarization

Automatically label different speakers in multi-party conversations with high accuracy. (Coming soon)

Multiple Language Support

Transcribe in multi-languages and dialects with accent robust models trained on diverse datasets. (Coming soon)

Performance

Zoom Speech ranks among the top models on the Open ASR Leaderboard

We're proud to share that Zoom's Speech Recognition technology ranks among the top models on the Open ASR Leaderboard, a global benchmark for automatic speech recognition (ASR) performance. This milestone reflects our relentless pursuit of excellence in speech technology and the strength of Zoom AI Services' Scribe API.
Key model strengths include:

Best-in-class Accuracy: Our scalable innovation framework continuously enhances model quality, providing best-in-class transcription results where accuracy and readability matter most.
Mastery of Enterprise Terminologies: Optimized for business and technical contexts, the model accurately handles company names, product terms, and domain-specific jargon — a critical advantage for meetings, support calls, and professional documentation.
Reduced Hallucinations: Zoom’s advanced modeling strategies minimize transcription “hallucinations” so what’s recognized truly reflects the speaker’s intent, not artificial or extraneous words.