Voicemaker API for developers

Build production-ready audio AI in minutes

APIs trusted by teams shipping at scale, including Text to Speech, Speech to Text, Voice Agents, and Music.

Get API key Explore Docs

Text to Speech API

Convert text to speech using AI voices

Replace YOUR_API_KEY with a key from your API dashboard. Keep it on your server, never in browser code.

  • Marvel
  • Netflix
  • TCS
  • Coca-Cola
  • Sony
  • Disney+
  • Amazon
  • Reddit
  • Flipkart
  • PhonePe
  • Infosys
  • DHL Express
  • Honeywell
  • Payoneer
  • HSBC
  • Tesla
  • Capgemini
  • Siemens
  • University of California
  • Harvard University
  • Pentagram
  • Finnair
  • United Airlines
  • Delhivery
  • TVS Motor
  • ExxonMobil Chemical
  • Atomberg
  • Emerson

Voicemaker APIs

SuperTTSAds, stories, and music scores from one prompt. API DocsEndpoints, parameters, and sample requests. Help CenterGuides for the editor, voices, credits, and billing. StatusLive availability for the API and the editor.

Pick the voice model for the job

Every model uses the same endpoints and request fields. Switch by changing the VoiceId.

Model Best for Languages Credits / character
ProPlus Expressive V2Our most advanced model, with prompt-based emotion and delivery Storytelling, audiobooks, premium voiceovers 90+ 4×
ProPlus Expressive V1Prompt-driven voices with deep emotional range Storytelling, characters, dubbing 70+ 4×
ProPlus High-ResStudio-grade clarity and fidelity Ads, broadcast, media production 30+ 4×
ProPlus TurboLow latency, about 250–300 ms Voice agents, chatbots, live dialogue 30+ 2×
Aura PrimeProduction-quality multilingual speech Narration, e-learning, product audio 94 1×
Aura FlashFastest, lowest-cost speech, same languages as Aura Prime High-volume and real-time workloads 94 0.5×
FlashXExperimental HD multilingual model Multilingual voiceovers 20 1×
Pro2Next-gen multilingual, strong on Indian languages Regional and multilingual content 30+ 2×
Pro1Standard neural multilingual voices Everyday multilingual narration 90+ 1×
Default voicesAI1–AI6 and HashCode Bulk TTS and high-volume workloads 130+ 1×
Output formats MP3WAVOGGOPUSAACμ-law / a-law 8 kHz

Real-time over WebSockets

For voice agents and live captions. Authenticate once when the socket opens, then stream.

Streaming Text to Speech

Open one socket and send a JSON message per line of text. Audio comes back as base64 chunks in order, and the last one carries isFinal: true. Pair it with ProPlus Turbo voices for live agents.

wss://developer.voicemaker.in/api/v1/voice/convert
  • Up to 3,000 characters per message
  • Text under 400 characters returns one chunk
  • mp3, wav, ogg, opus, aac, μ-law, a-law
{
  "VoiceId": "ai3-Jony",
  "LanguageCode": "en-US",
  "Text": "Welcome to the Voicemaker API.",
  "OutputFormat": "mp3",
  "SampleRate": "48000"
}
Read the WebSocket guide

Real-time Speech to Text

Stream microphone or call audio and get partial transcripts while the speaker talks. A phrase is committed after a short silence, or when you send a commit yourself.

wss://developer.voicemaker.in/api/v1/speech-to-text/realtime
  • 16-bit PCM from 8 to 48 kHz, or μ-law 8 kHz
  • Voice-activity or manual commits
  • Optional word timestamps
{
  "type": "start",
  "model": "stt-flagship-realtime",
  "audioFormat": "pcm_16000",
  "language": "en-US",
  "commitStrategy": "vad"
}
Read the WebSocket guide

API reference at a glance

Base URL https://developer.voicemaker.in. Every request needs an Authorization: Bearer header.

Text to Speech

Speech to Text

Voice Cloning

Speech to Speech and VoxFX

Trusted by teams building on the Voicemaker API

Localization pipeline

Our localization pipeline calls the convert endpoint for every market as soon as a trailer script is approved. Guide tracks land in the edit the same hour, and only the lines that change are regenerated.

LMS integration

Our course platform sends each edited paragraph to the API and stores the returned audio next to the lesson. Quarterly compliance updates publish in every region without anyone opening an audio editor.

Telephony & IVR

Delivery prompts come back as 8 kHz μ-law, so they drop straight into the phone system. New menus ship from our release pipeline in several languages, all in one approved voice.

Captions API

Lecture recordings go to Speech to Text as soon as they are uploaded. Speaker labels and SRT files come back with the transcript, so every video has captions before students open it.

Batch generation

Cabin and gate announcements are generated in batch for every route and language. When a procedure changes, one request regenerates that line in the same voice and the next build picks it up.

Regional voice-overs

Sale-season scripts fan out to Hindi, Tamil, Telugu and more through one API key. Pro2 voices handle the regional languages, and the full set of voice-overs is ready in an afternoon.

Banking support

Phone banking prompts are generated from signed-off scripts through the API, so the audio always matches the approved text. A wording change is one request, not a studio session.

Games & media

Writers call the API from our dialogue tool to hear scenes with expressive ProPlus voices while the script is still changing. VoxFX presets add radio and creature effects to the temp tracks.

Enterprise security with fine-grained controls

ISO 27001AICPA SOC 2FedRAMP High
Inputs and outputs are always your IP
Every session runs in its own VM
Zero data retention enforced with model providers
RBAC / SSO / SCIM / OIDC
Comprehensive audit logs
Private networking available

Frequently asked questions

Authentication, limits, formats, and billing for the Voicemaker API.

Send your API key as a bearer token in the Authorization header: "Authorization: Bearer YOUR_API_KEY". Create keys in the API keys dashboard on developer.voicemaker.in, and keep them on your server, not in browser or app code.

The REST convert endpoint takes up to 10,000 characters per request; requests over 3,000 characters can take longer. The Text to Speech WebSocket and VoxFX take up to 3,000 characters per message.

By default, the convert endpoint returns a URL to the audio file, kept for 1 hour. Set FileStore to keep it for up to 240 hours, or set ResponseType to "stream" to receive the audio bytes directly.

Text to Speech outputs mp3, wav, ogg, opus, aac, and μ-law or a-law for telephony, at sample rates from 8,000 to 48,000 Hz. 48,000 Hz gives the best quality, and mp3 gives the lowest latency.

Text to Speech is charged per character: 1× for default and Pro1 voices, 2× for Pro2 and ProPlus Turbo, and 4× for ProPlus Expressive and High-Res. Chinese, Japanese, and Korean text counts 2× on default and Pro1 voices. Speech to Speech and Speech to Text are charged per second of audio.

Files of 3 minutes or less are transcribed in the same request. Longer files are processed in the background: the response returns a taskId, and you fetch the result from GET /api/v1/speech-to-text/{taskId}. Jobs stay listed for 1 hour.

Speech to Speech takes ProPlus voices and your own cloned voices. Upload mp3, wav, or mp4; the timing and pacing of the original recording are kept.

A paid plan with an unused voice clone slot, and 20 seconds to 30 minutes of audio in up to 20 mp3, wav, or mp4 files (200 MB in total). Cloning runs in the background, and the new voice appears in your clone list when it is ready.

Yes. AI1, AI2, AI3, and AI4 voices accept inline SSML tags such as emphasis, break, and prosody in the Text field.

The most realistic audio AI platform

Explore Docs Get API key
Saving changes