Transcribe an audio file from a given URL. The API will download the audio found at the URL and transcribe it at our crowdsourcing platform.
JSON Parameters
Parameter |
Example |
Required |
Description |
audioSrcUrl |
Yes |
URL for the audio file to be transcribed. This must be a publicly available URL or a temporarily public URL. |
|
uuidProject |
7cf22963-51d2-4369-8030-bf6df57daf8a |
Yes |
UUID of the audio transcription project. |
tags |
["test26", "test1"] |
No |
Tags can be used to add context to transcription or keep track of certain topics in your pipeline. Use it to categorize, prioritize, or further diagnose transcription progress based on your projects or customers' needs. |
speakerCount |
n |
No |
The number of speakers in the audio. If transcribing dual channel (stereo) audio, and you know there is exactly one speaker on each channel (for example, a phone call), setting speakerCount to 2 will enable speaker labels. If unsure of speaker count, make an uppermost guess. |
languageCode |
en |
Yes |
The language code for the language used in the input media file. Can be en, es, etc. |
locale |
US |
Yes |
Country code related to accent. |
options |
{"formatText": false} {"truncatedWords": false} {"overlpadedVoices": false} {"nonAudible": false} {"nonVerbal": false} {"phoneticWriting": false} {"interjections": false} |
No |
Optional parameters for manipulating the transcript. Setting formatText to false will disable punctuation and casing. Setting truncatedWords to false will disable annotation of truncated words. Setting overlapedVoices to false will disable annotation of voice overlapping. Setting nonAudible to false will disable annotation of nonaudible speech fragments. Setting nonVerbal to false will disable annotation of nonverbal human noises like laughs. Setting phoneticWriting to false will disable annotation of words written according to their phonetics. Setting interjections to false will disable transcription of Uhms and Ers. |
Example: Create a transcript and disable text formatting and truncated words annotation.
curl --request POST \
--url 'https://api.atext.io/v1/transcripts' \
--header 'Authorization: Basic AUTHORIZATION_TOKEN' \
--header 'Content-Type: application/json'\
--data '
{
"audioSrcUrl": "AUDIO_URL",
"projectUuid": "PROJECT_UUID",
"tags": ["sample", "demo-1"],
"speakerCount": 2,
"languageCode": "es",
"locale": "US",
"options":
{
"formatText": false,
"truncatedWords": false
}
}'
Expected Response
{
"uuid": "16a54748-da46-4080-b4a1-fa18a6b223cf",
"status": "started",
"audioSrcUrl": "AUDIO_URL",
"language": "es",
"locale": "US",
"tags": [
"sample",
"demo-1"
],
"speakerCount": 2,
"options": {
"formatText": false,
"truncatedWords": false
},
"createdAt": "2019-12-08T01:25:49.031Z"
}
When you create a transcript, the status goes from started to processing to ready. Processing time normally takes under 24 hours.
To get the result of the transcript, pull for the transcript id with GET requests.