Audio

Overview

AGTCloud audio APIs support speech synthesis, audio transcription, and audio translation. Each task uses a different input format and returns either audio bytes or text.

Supported tasks

Task	Description	Main result
Speech synthesis	Convert text into speech	Binary audio
Audio transcription	Convert audio into text in the source language	JSON, text, or subtitles
Audio translation	Convert audio into English text	JSON, text, or subtitles

API access

Base URL

https://api.agtcloud.cn/v1

Speech synthesis

curl 'https://api.agtcloud.cn/v1/audio/speech' \
  -H 'Authorization: Bearer your-api-key' \
  -H 'Content-Type: application/json' \
  --data '{
    "model": "MODEL_ID",
    "input": "Welcome to AGTCloud.",
    "voice": "alloy"
  }' \
  --output speech.mp3

Parameter	Required	Description
`model`	Yes	Text-to-speech model ID
`input`	Yes	Text to synthesize
`voice`	Yes	Voice used for generated speech
`response_format`	No	Audio output format
`speed`	No	Speech speed

Speech returns audio bytes. Save them using the actual response format instead of parsing the response as JSON.

Audio transcription

curl 'https://api.agtcloud.cn/v1/audio/transcriptions' \
  -H 'Authorization: Bearer your-api-key' \
  -F 'file=@meeting.mp3' \
  -F 'model=MODEL_ID'

Parameter	Required	Description
`file`	Yes	Audio file to transcribe
`model`	Yes	Transcription model ID
`language`	No	ISO-639-1 language code for the input audio
`response_format`	No	Output format for the transcript

Transcriptions uploads audio as multipart data. In the basic JSON response, the transcript is in text.

Audio translation

curl 'https://api.agtcloud.cn/v1/audio/translations' \
  -H 'Authorization: Bearer your-api-key' \
  -F 'file=@chinese.mp3' \
  -F 'model=MODEL_ID'

Parameter	Required	Description
`file`	Yes	Non-English audio file to translate
`model`	Yes	Audio translation model ID
`prompt`	No	English text that guides output style
`response_format`	No	Output format for the translation

Translations returns English text. In the basic JSON response, the result is in text.

Handle responses

API	Handling
Speech	Save binary audio using the response format
Transcriptions	Process JSON, text, or subtitles according to `response_format`
Translations	Process the English text according to `response_format`