Automated transcription is a part of the workflow for professionals across many industries. Several content creation companies, developers, video platforms, podcasting platforms, and many more use speech-to-text API’s to transcribe & caption audio and video.
Regardless of your industry, there are a few important factors you’ll want to consider when selecting the best speech-to-text service for your needs. Price, security, speed, and accuracy are the four factors that differentiate one service from the next.
With more options for speech-to-text AI services popping up all the time, the easiest way to determine the right one for your needs is to compare two leading services side by side. We’re taking a look at Amazon Transcribe vs. Rev AI to see how they measure up against each other.
Comparing Rev AI to Amazon Transcribe
To help you understand how Rev AI works and how it compares to Amazon Transcribe, we’re going to look at five specific areas:
- Price—What is the cost per audio minute to convert audio to text?
- Accuracy—How accurately is the audio converted to text? In automatic transcription, this is also referred to as Word Error Rate (WER). WER determines the accuracy of the text. A low rate indicates that the transcription closely matches the audio.
- Security—Are the files I submit kept safe?
- Speed—How fast is the audio I submit converted into text and delivered back to me?
- Maximum file length & limits – What is the maximum file length you can upload to the service? How many calls can you make?
Amazon Transcribe vs. Rev AI
1. Price per minute
Prices on these platforms change all of the time and vary depending on order volume & packages, but you can see the Amazon Transcribe prices here and the Rev AI prices here.
As of when this article was written, Rev’s Enterprise package is $1.20 per hour, or $0.02 per minute.
For the first 250,000 minutes, Amazon Transcribe is $0.024 per minute, and it gets cheaper per minute when you get above 750,000 minutes.
Rev AI
Amazon Transcribe
2. Speech to text accuracy (as of late 2020)
Rev has a much lower word error rate (WER) than Amazon Transcribe according to several tests, both tests done by Rev and tests done by third parties. When measuring word error rate, the lower number is more accurate (less errors).
Rev: 14.22% word error rate (WER)
Amazon Transcribe: 18.42% word error rate (WER)
3. Secure audio files
Rev: TLS 1.2 Data Encryption and Secure Servers protect your data. Rev has achieved SOC2 Type 1 Security Compliance. Read the full security & privacy program overview here and learn more on the Rev AI security & privacy landing page.
Amazon Transcribe: Read the full privacy information in Amazon Transcribe’s Privacy Policy. You can also see data privacy FAQ’s here
4. Speed and turnaround time
Rev: typically about 5 minutes
Amazon Transcribe: typically about 8 minutes
5. Maximum file length & limits
Rev: For the async API, it is a 17 hour maximum for English transcription files and 12 hour maximum for other languages offered. You can also submit 10,000 transcription requests every 10 minutes, and 500 transcriptions can be processed every 10 minutes.
For the Rev AI streaming API, the time limit per stream is 3 hours.
Amazon Transcribe: API calls are limited to 4 hours (or 2GB) per API call for the batch service. The streaming service can accommodate open connections up to 4 hours long.
Testing a New Transcription Service or API
Before sending a bulk order to a new speech to text API, we recommend testing it out first. Rev AI provides a free API trial here for your first 5 hours of audio or video content.
You can also test out speech recognition WER (word error rate) & accuracy yourself with our new FST Align developer tools. Test out Rev AI vs. Amazon and other competitors to see who has the most accurate speech-to-text AI.
