Cloud Speech-to-Text

What it does

Cloud Speech-to-Text takes audio as input and returns text transcriptions. Audio can be sent inline, referenced from Cloud Storage with a gs:// URI, or streamed over gRPC from a microphone or live source. Results come back as text alternatives with optional confidence values, word-level details and speaker labels, depending on the features enabled in the request.

Developers typically call the API from their own applications using client libraries, choosing between synchronous recognition for clips up to one minute, asynchronous recognition for files up to 480 minutes, and streaming for live audio. Model selection, including the Chirp 3 transcription model, custom word hints and domain-specific models such as phone call or video, shapes accuracy. The console also offers a no-code uploader for quick tests before integration.

The service suits developers and enterprise teams building transcription, captioning or voice-control features who are comfortable with Google Cloud billing and projects. Access requires a Google Cloud account, a project with billing and the Speech-to-Text API enabled, and a per-minute charge for processed audio. Confidence values may not always be set and should not be treated as required fields, and audio must be encoded in supported formats.

What it can do

• Transcribe short audio up to one minute with synchronous recognition
• Transcribe long audio up to 480 minutes with asynchronous, polled recognition
• Return real-time interim results from streaming audio over gRPC
• Use model adaptation and hints to favor domain words and rare terms
• Separate speakers with diarization and recognize multiple audio channels
• Filter profanity and apply automatic punctuation (beta)

Works with

• Google Cloud Storage (audio referenced by gs:// URI)
• Google Cloud console uploader and in-console tutorials

Example workflows

Transcribe a recorded meeting file

Input: a multi-hour audio file stored in Cloud Storage. Steps: submit an asynchronous recognition request with the gs:// URI and a language code, enable speaker diarization, then poll the long-running operation until results are ready. Output: a transcript split into utterances with speaker labels.

Live captions from a microphone

Input: a live audio stream from an application microphone. Steps: open a gRPC streaming recognition session with a LINEAR16 or supported encoding at 16000 Hz, display interim results as they arrive, and replace them with final results marked is_final. Output: real-time on-screen captions.

Before you choose

• Asynchronous recognition supports up to 480 minutes, while synchronous requests are limited to 1 minute of audio.
• Charges accrue per minute of successfully processed audio, multiplied per channel, so multichannel audio costs more.
• Do not require confidence values in code, since they may be unset or inaccurate.

Listed under

• Coding
• API
• Audio
• Call transcription
• Captions
• Developers
• Enterprise

Get started

Sign in to the Google Cloud console and enable the Speech-to-Text API for a project.

1. Create or select a Google Cloud project, sign in to the console, and enable billing for it.
2. Enable the Speech-to-Text API for the project and set up credentials for your client library.
3. Choose a sync, async or streaming method and set the languageCode, encoding and sample rate, using 16000 Hz where your source allows it.

New customers can receive up to $300 in free Google Cloud credits, which is a time-limited trial allowance rather than a permanent free tier. Standard V2 recognition is charged per minute with volume tiers; dynamic batch is discounted. Multi-channel audio bills per channel, and any Cloud Storage or other Google Cloud usage is billed separately. Check the pricing page for current rate tables.

Source: cloud.google.com/speech-to-text/pricing, checked 28 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 28 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Speech-to-Text: AI voice typing & transcription | Google Cloud ↗cloud.google.com
• Speech-to-Text API Pricing | Google Cloud ↗cloud.google.com
• Cloud Speech-to-Text overview  |  Google Cloud Documentation ↗docs.cloud.google.com

Company website

cloud.google.com: product, pricing and policy pages

YC directory mirror

Batch and company name

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 28 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.