Transcribe

Amazon Transcribe is an automatic speech recognition (ASR) service, and offers various features which improve this process: language customisation, filters for privacy, audience appropriate language, as well as speaker identification. The language can either be specifically set or set to be autodetected. Transcribe’s output is full-text searchable text, meaning that it also performs indexing of the produced content.

You can also configure custom vocabularies and create custom language models.

Transcribe comes in a number of specific flavors:

  • Transcribe

  • Transcribe Medical: for conversations among medical personnel

  • Transcribe Call Analytics: for categories and sentiment of people on phone calls

Features include:

  • Language autodetection

  • Speaker identification: useful when working on meetings recordings or subtitles and captions.

  • Filtering for PII and other redactions

  • Custom vocabularies

  • Partial result stabilization: Because streaming works in real time, transcripts are produced in partial results. Amazon Transcribe breaks up the incoming audio stream based on natural speech segments, such as a change in speaker or a pause in the audio. Approximation is the process of returning the chunk with a partially converted ending: While saying "The Amazon is the largest rainforest on the planet", you might get partial results along the way such as:

"Law", "ray", "rain" appear because you haven’t completed pronouncing the word.
The
The Amazon.
The Amazon is
The Amazon is the law.
The Amazon is the largest
The Amazon is the largest ray
The Amazon is the largest rain for
The Amazon is the largest rainforest.
The Amazon is the largest rainforest on the
The Amazon is the largest rainforest on the planet.

You can use Transcribe via the console UI and APIs for integrating it with your application, it also integrates with other machine learning services.

Transcribe is pay-per-use and your billed per second of transcribed audio.