Clone a voice from audio
Build narration workflows with a reference recording and a script. Use the AI Voice Clone API when your feature needs speech based on an existing voice.
AI Audio APIs
Integrate voice cloning, voice design, and music generation into your app. Compare the audio and text inputs each API requires, then choose the endpoints for your audio workflow.
Find your API ↓
Choose APIs for your audio workflow
Design a voice from text
Add speech generation from a voice description and script. Use the AI Voice Design API to define a speaker without a reference recording.
Generate music from a prompt
Integrate music generation using a text prompt for mood and style. Configure the instrumental option to match the audio your workflow needs.
3 API products
AUDIO
AI Voice Clone API for Reference-Based Speech
Build voice-based narration into your app using a reference recording and a new script. Submit both inputs and retrieve the generated speech as an audio URL for your workflow.
1 endpoint
AUDIO
AI Voice Design API for Described Speech
Add custom voice generation to your app with a script and a text description of the voice. Retrieve the generated speech without supplying a reference recording.
1 endpoint
AUDIO
AI Text to Music API for Musical Briefs
Integrate music generation into your app using a text description. Supply an optional title, set the instrumental flag, and retrieve the generated track as an audio URL.
1 endpoint
What makes VidMage's AI Audio APIs stand out
Give your app three ways to create audio: a voice from a recording, a voice from a written brief, or music from a prompt. Each returns an audio URL through the same task workflow.
Turn a reference voice into a new script
Voice Clone takes a reference recording and up to 1,000 characters of new speech in textContent. Keep the recording and script separate so users can create another narration from the same voice reference. Validate script length before submitting.
Learn more ↗Create a narrator without a recording
Voice Design accepts a voice description and a script of up to 500 characters in textContent. Your users can describe a calm guide or an energetic character without uploading a speaker sample. Validate the script against this API's limit.
Learn more ↗Ask for an instrumental with one setting
Text to Music accepts a description of 1 to 400 characters, an optional title of up to 80, and a makeInstrumental boolean. Validate these limits and expose the instrumental choice beside the prompt so users can request a backing track for a montage or voiceover.
Learn more ↗One API key for code and AI clients
Use the same VidMage API key to call audio tools through REST or an MCP client. All three audio APIs return a completed audioUrl, so your app can share task tracking and a player across speech and music.
Learn more ↗FAQs
Answers for developers building with AI Audio APIs.
Does voice cloning create a voice ID I can reuse?
The published API does not expose voice enrollment or a permanent voice ID. Send the reference recording with each request. Keep your recording and request history in your app to organize recurring narration work.
Can I set music duration, BPM, or separate stems?
There are no dedicated duration, BPM, lyrics, or stem controls in the current music request. Describe the musical direction in the prompt. The API returns one audio file rather than separate stems.
How do I upload a local voice recording?
Create a temporary upload for voice-clone and its audioUrl parameter, upload the file bytes with the returned instructions, then submit fileUrl as audioUrl. Check the current upload rules for accepted formats and sizes; the voice request schema does not publish a complete audio-format table.
Does the API stream audio as it is generated?
The published contract has no audio streaming interface. Show a processing state while you query the submitted task, then play the completed result. Keep taskId if the user leaves or your local wait ends.
How are audio API requests billed?
API access requires an active subscription, and API tasks use the same credit balance as website tasks. Use the read-only MCP capability estimator to check your exact request without spending credits. Keep the final task usage record as well as the estimate.
Can I rely on these APIs for production audio workloads?
VidMage's Developer APIs are currently in beta and are not yet recommended for production. Use them to test your audio workflow, input validation, task recovery, and output quality. Recheck the current availability notice and API contract before planning a production launch.
Will a failed or timed-out audio request be refunded?
Confirmed failures are refunded automatically only when the refund outcome is certain. A timeout does not confirm failure. Keep the task ID and original idempotency key, check the existing task or recover recent tasks, and follow the returned recovery instructions before submitting again.