Sync Speech-to-Text API

Send an audio file in a single HTTP request, get a transcript back in milliseconds. No polling, no session management.

Record or and get back an accurate transcript instantly...

import requests

with open("sample.wav", "rb") as f:
    audio = f.read()

response = requests.post(
    "https://sync.assemblyai.com/transcribe",
    headers={
        "Authorization": API_KEY,
        "X-AAI-Model": "universal-3-5-pro",
    },
    files={"audio": ("sample.wav", audio, "audio/wav")},
)

result = response.json()
print(result["text"], result["confidence"])

Model

One call, one finished transcript

Send a short clip and receive a transcript almost instantly. Universal-3.5 Pro accuracy, returned in ~134 ms.

Sync API

Universal-3.5 Pro accuracy in a single request/response call.

$0.45 / hr

Start building

Best for: dictation, voice agents with their own turn detection, short clips
Up to 2 minutes of audio per request (max 40 MB)
~134 ms p50 to a finished transcript
Per-word text, timing, and confidence in every response
Keyterms prompting and conversation context included

Compare

Process your audio now, as it happens, or later

Model

Sync Speech-to-Text API One request → one response

Realtime Speech-to-Text API Live, ongoing session

Pre-recorded Speech-to-Text API Submit → poll or webhook

Transport

HTTP POST

WebSocket

HTTP POST + poll/webhook

Transcript arrives

In the same response

While the audio streams

When the job completes

Best for

Dictation, voice agents with your own turn detection, short clips

Live captions, real-time voice agents

Long audio, batch, latency-tolerant workloads

Audio length

Up to 2 minutes per request

Unbounded live sessions

Long-form files

Integration

One POST handles the entire request

Manage a WebSocket session

Job lifecycle with polling or webhooks

Speaker diarization

—

PII redaction & speech understanding

—

Price

$0.45 /hr

From $0.21 /hr

Learn more

Read the docs

Learn more

Note: The Sync API does not include PII redaction, speaker diarization, or speech understanding — if you need these features, use the Pre-recorded or Realtime APIs.

Use cases

Built for bounded audio

Anywhere the user controls the start and stop of a short clip and needs the words back immediately.

Dictation

Speak, see it typed. A single round trip at ~160 ms average latency makes dictation and voice input feel instantaneous — with accuracy that on-device models can't match.

Voice agents

Run your own turn detection? You detect the turn, we transcribe it. Extremely fast, accurate ASR in a single call.

Push-to-talk

Mobile and desktop voice commands captured in a single request. Reliable over poor connections where persistent sockets would fail.

Voicemail

Voicemails and short-form clips under 2 minutes get a faster home than async. One call returns the transcript instantly — no job ID to check, no status polling.

IVR

Transcribe one utterance, route the call, move on.

Edge functions

One stateless HTTP call is built for edge functions, serverless workloads, and environments where persistent sockets aren't an option.

Playground

Test the API in seconds

Put the Sync API to the test with your own audio in our no-code playground.

Try it out

Common questions

: The Sync API is a synchronous, request/response transcription endpoint. You send a short audio clip (up to 2 minutes) in a single HTTP request and get a finished transcript back in the same response — typically in ~134 ms at p50. There is no job to submit, no status to poll, no webhook to catch, and no WebSocket to manage.
: Every transcription workload comes down to whether you need the answer fast, soon, or eventually. The Realtime Speech-to-Text API returns words live over a WebSocket while audio is still arriving. The Pre-recorded Speech-to-Text API submits a job for long files and returns results via polling or webhook. Sync is the third path: one short clip in, one finished transcript back, in a single round trip.
: The Sync API is powered by Universal-3.5 Pro — the same flagship model behind our Pre-recorded and Realtime products, not a stripped-down fast variant.
: Sync transcription is $0.45 per hour of audio — the same rate as Universal-3.5 Pro Realtime. Keyterms prompting and conversation context are included at no extra cost. There are no rate limits and no upfront commitments, and volume discounts are available.
: Up to 2 minutes of audio per request, with a 40 MB maximum. Because latency is predictable from input length, oversized requests are rejected upfront and pointed to the Pre-recorded Speech-to-Text API rather than failing partway through.
: The Sync API is purpose-built for short, bounded audio and focuses on accuracy and speed. For PII redaction, speaker diarization, and speech understanding, use the Pre-recorded or Realtime APIs.
: The Sync API supports the same 18 languages as Universal-3.5 Pro — English, Spanish, French, German, Italian, Portuguese, Arabic, Danish, Dutch, Hebrew, Hindi, Japanese, Mandarin, Vietnamese, Finnish, Norwegian, Swedish, and Turkish — with prompt steering. It defaults to English.
: Create a free account and use your existing API key — send a short audio clip and read the transcript off the response. You can also try it with your own audio in the Playground, no code required.