Speech-to-Text Benchmark Report

See how our accuracy compares to your current provider, and other APIs like Google, and AWS.

How it Works

Submit your files

Share 5-10 audio/video files with us that we'll use to conduct the accuracy benchmark.

We run the benchmarks

Files will be human-transcribed to 100% accuracy an compared to our transcription results - along with your current API, Google, and AWS.

Review Benchmark Report

Directly compare our accuracy, pricing, features, and support. Raw transcripts will be provided.

AssemblyAI Benchmark Report.gif

What's in the Report

Compare our API side-by-side with your current provider


Review the accuracy %, or Word Error Rate (WER), for each of the files you share.


Compare costs across providers depending on your volume and speed requirements.


Analyze our full feature set and how they can be used within your product.

Start Your Benchmark Report

Please fill out the following form and someone from our team will be in touch about next steps


Is the benchmark report free?


Why should I get a benchmark report?

Shopping for speech-to-text can be tricky. Every provider claims they have the "best" accuracy. Instead of promising, we'd like to show you first hand how well we perform. We provide all raw transcripts in case you'd like to run any additional internal benchmarking.

Do you share or reuse any of my audio or video data?

No. All audio and video files uploaded are never stored and are deleted upon processing.

How is accuracy, using the Word Error Rate (WER), calculated?

Here are the steps we take to calculate accuracy:
1 Transcribe files automatically through APIs (AssemblyAI, 'Your Provider', Google, & AWS)
2 Transcribe files manually to 100% accuracy
3 Compare API with human transcripts to calculate Word Error Rate (WER)

What is the Word Error Rate (WER)?

WER is the industry-standard metric to test transcription accuracy. It calculates the total errors in the automated transcription versus human transcription (100% accuracy).Example: WER of .10 = 90% accuracy