Skip to main content

Overview

An overview of the speech assessment API, its capabilities and integration workflow

The speech assessment API (SAAPI for short) returns a detailed speech report for a recording of English speech.

The API has three main versions:

  • Pronunciation: reading (prompted) assessment of spoken English for pronunciation only.
  • Scripted: reading (prompted) assessment of spoken English for pronunciation and reading fluency.
  • Unscripted: speaking (spontaneous) assessment of spoken English where the response is unknown

Pronunciation & Scripted API​

Integration workflow​

For a scripted API request, you send us a recording of one of your users speaking + the expected text. Our API returns a detailed report with scores, metrics and feedback which you can then use to display back to your users via your own UI front-end.

Unscripted API​

Integration workflow​

For an unscripted API request, you only send us the recording of one of your users speaking since the content they may speak is open-ended and unscripted. Our API returns a detailed report with scores, metrics and feedback which you can then use to display back to your users via your own UI front-end.

Comparison of features​

FeatureDescriptionPronunciationScripted APIUnscripted API
Pronunciation scoringReturns a pronunciation report with scores at the phoneme, word and overall level✅✅✅
Fluency scoringReturns a fluency report with score, metrics and feedback✅✅
Reading Assessment MetricsReturns reading metrics✅
Return predicted textReturns the predicted text detected by the speech recognition✅✅
Scripted content relevanceReturns how close the user's speech was to the expected content✅
Unscripted content relevanceReturns how relevant the users answer was to the question asked.✅
Grammar scoringReturns a grammar report with score, metrics and feedback✅
Vocabulary scoringReturns a vocabulary report with score, metrics and feedback✅
English test scoring estimatesReturns estimated predictions for IELTS, PTE and CEFR✅✅✅