Advanced Speech-to-Text that just works.

Use the AssemblyAI API to transcribe audio from phone calls, podcasts, and all types of recordings. Start building today, no sales calls or contracts required.

import assemblyai

aai = assemblyai.Client(token='your-api-token')

# Optional: boost accuracy for keywords/phrases
phrases = ['cancel my account', 'Yiqin Dai', ...]
model = aai.train(phrases)

# Transcribe audio in any format
url = 'https://foo.com/bar.mp3'
transcript = aai.transcribe(audio_url=url, model=model)

Why choose AssemblyAI?

AssemblyAI uses advanced deep learning technology to generate extremely accurate transcriptions for your audio. See how we compare to services like Google, VoiceBase, and AWS Transcribe.

AssemblyAI

oh i'd say there's no such thing as eating too much but i just haven't massively fast metabolism and so i constantly eating so

Google (phone call model)

i see this message thing is eating too much but i just haven't massively fast metabolism and snow i constantly ding

Completely customizable to your application

Easily boost accuracy for keywords or phrases that are important, or add thousands of custom words to the vocabulary, to fine-tune the recognition for your specific needs.

Boost Accuracy for Keywords/Phrases

Add Custom Vocabulary

import assemblyai

aai = assemblyai.Client(token='your-api-token')

# Boost accuracy for an unlimited amount of important
# keywords/phrases by creating a custom model
# Models take around 6 minutes to train.
phrases = ['cancel my account', 'Dirk Gently', ...]
model = aai.train(phrases)

# Transcribe audio in any format
url = 'https://foo.com/bar.mp3'
transcript = aai.transcribe(audio_url=url, model=model)

text = transcript.text

{
"transcript": {
  "id": 40,
  "status": "completed",
  "created": "2017-11-12T05:00:05.113353Z",
  "audio_src_url": "https://foo.com/bar.wav",
  "model_id": null,
  "text": "Welcome to AssemblyAI.",
  "confidence": 0.98,
  "segments": [...],
  "speaker_count": null
 }
}

More Reasons to Choose AssemblyAI

continuously improving
Secure and Private

We believe in privacy. We never store, copy, or share the audio data you send to our secure API. Your audio data is deleted from our servers immediately after our algorithms transcribe it. Contact us to run the API on your own servers.

Custom Models
Custom Models

Use the API to boost accuracy for an unlimited number of important keywords or phrases, or to add thousands of custom terms to the vocabulary.

continuously improving
Supports All Audio Formats

The API accepts virtually any audio format, even lossy and low bitrate audio commonly found in phone calls. No need to worry about sample rates, bit rates, encodings, or other tricky signal processing terminologies.

continuously improving
More Affordable

Pricing is a simple $0.0003 per second of audio sent to the API, billed monthly, without any weird rounding or minimums.

Ready to get started?