Score Text/Phoneme list

circle-info

With the phone list API, you can score a sequence of phonemes that make up any word or non-word.

Individual words, such as "Gotcha," which are American vernacular and not valid dictionary words, can be scored using the phoneme list API. The phoneme list uses a different url endpoint and expects the list of phonemes in Arpabet notationarrow-up-right.

Request Format

The endpoint that is to be used will depend on the region of your subscription. For example, for US West, the endpoint is https://api.speechace.coarrow-up-right.

POST https://api.speechace.co/api/scoring/phone_list/v9/jsonarrow-up-right

curl --location -g 'https://api.speechace.co/api/scoring/phone_list/v9/json?key={{speechacekey}}&dialect=en-us' \
--form 'phone_list="g|ao|ch|ah"' \
--form 'user_audio_file=@"gotcha.wav"'

Query Parameters

Parameter
Type
Description

key

String

API key issued by Speechace.

dialect

String

This is the dialect to be scored. Supported only for en-us and en-gb

user_id

String

Optional: A unique anonymized identifier for the end-user who spoke the audio.

Request Body

Parameter
Type
Description

phone_list

String

A phoneme list to score. For example: g|ao|ch|ah

user_audio_file

File

file with user audio (wav, mp3, m4a, webm, ogg, aiff).

question_info

String

Optional flag: A unique identifier (generated by your application) for the activity or question this user audio is answering.

Response Example

The API response includes phone_score_list[] : a list of phonemes with it's own quality_score.

Each element within the phone_score_list[] also includes its own quality_score, its extent information indicating its start and end, and additional fields like the actual sound_most_like phone based on the speaker's attempt.

Last updated