Speech synthesis
29 voices, emotions, speed, and pitch. MP3, WAV, OGG, and chunk streaming for long texts.
TTS · STT · Realtime · Intelligence
Neural synthesis and voice cloning with no limits — create as many of your own voices as you like from a single recording. Plus streaming transcription, VAD turn detection, and intelligence analysis. REST and WebSocket — from one request to live sessions.
Features
Synthesis, recognition, and an intelligence layer over audio — in one API.
29 voices, emotions, speed, and pitch. MP3, WAV, OGG, and chunk streaming for long texts.
Your own voice from one recording: 3–30 seconds of clean speech and synthesis in 9 languages.
Audio to text with word-level timestamps, keyterm prompting, and automatic language detection. Up to 15 minutes.
Real-time transcription over WebSocket: partial and final events with VAD endpointing.
Finds speech boundaries: speech_started / speech_ended events — the basis for voice agents.
Who is speaking: splitting a recording into speakers for meetings, interviews, and call centers.
Speech mood by segment: joy, anger, neutral. Useful for HR, marketing, and support.
Topics, summaries, PII redaction, and language detection — no audio upload required.
Automatic tags for archives, search, and SEO — extracted from transcripts effortlessly.
Reverb, EQ, compressor, pitch, and more — a chain of 7 effects for audio and video, applied as an async job.
Turn articles and books into speech: async synthesis with a sample-accurate WAV and a timed paragraph manifest.
Official Python and TypeScript SDKs for every endpoint.
Bulk synthesis and analysis in one request, processed in the background.
Notifications when async jobs are ready.
Full Swagger spec for client generation.
Demo
Up to 500 characters, no sign-up. Pick a voice and speed.
No-code
The dashboard includes a playground: synthesis, transcription, and analysis right in the browser — not a single line of code. Requests go out under your key and count toward your usage.
Plans
Start free. Top up your balance with crypto and pay for any plan in your dashboard.
Free / month
$4.90 / month
$19.90 / month
$99.90 / month
Some plans are temporarily unavailable while we scale up capacity — they'll be back soon.
Integration
Key in X-Api-Key or X-RapidAPI-Key. Synthesis returns binary audio; transcription, VAD, and analysis return JSON.
One POST — a ready audio file.
POST /v1/synthesize
X-Api-Key: rtt_…
Content-Type: application/json
{
"text": "Добрый день. Это синтез русской речи.",
"voice": "preset_anna",
"format": "mp3"
}Async job with a webhook: get job_id and fetch the result.
POST /v1/transcribe
X-Api-Key: rtt_…
Content-Type: multipart/form-data
audio=meeting.mp3
language=ru
diarization=true
→ 202 { "job_id": "…" }
GET /v1/transcribe/{job_id}
→ { "transcript": "…", "segments": […] }Upload audio and get speech segments back.
POST /v1/vad
X-Api-Key: rtt_…
Content-Type: multipart/form-data
audio=call.mp3
→ {
"segments": [{ "start": 1.2, "end": 4.8 }],
"speech_ratio": 0.42,
"duration_seconds": 8.6,
"processing_time_ms": 123
}Transcription plus emotions, keywords, entities, and speakers.
POST /v1/analyze
X-Api-Key: rtt_…
Content-Type: multipart/form-data
audio=call.mp3
language=ru
→ {
"transcript": "…",
"segments": […],
"keywords": […],
"entities": […],
"speakers": […]
}Async synthesis for long texts with a timed paragraph manifest.
POST /v1/synthesize/async
X-Api-Key: rtt_…
Content-Type: application/json
{
"text": "Глава 1. …\n\nГлава 2. …",
"voice": "preset_anna"
}
→ 202 { "job_id": "…" }
GET /v1/synthesize/async/{job_id}
→ { "paragraphs": [{ "start": 0.0, "end": 12.4 }] }
GET /v1/synthesize/async/{job_id}/audio
→ WAV · sample-accurateA chain of effects for audio and video — an async job with a webhook.
POST /v1/audio/effects
X-Api-Key: rtt_…
Content-Type: multipart/form-data
audio=voice.wav
effects=[{"type":"reverb"},{"type":"eq","bands":[…] }]
output_format=mp3
→ 202 { "job_id": "…" }
GET /v1/audio/effects/{job_id}/audio
→ processed audioWebSocket endpoints for live scenarios: streaming transcription and speech detection with events.
Send PCM16 frames — receive partials during speech and a final at the end of each phrase.
WS /v1/transcribe/stream?api_key=rtt_…
→ send: PCM16 · little-endian · mono · 16 kHz
→ send: {"type":"stop"}
← {"type":"session","event":"started"}
← {"type":"partial","text":"Привет, "}
← {"type":"final","text":"Привет, мир!"}
← {"type":"vad","event":"speech_ended"}
← {"type":"session","event":"ended"}Catch speech_started / speech_ended to drive dialogue and interruptions.
WS /v1/vad/stream?api_key=rtt_…
→ send: PCM16 · little-endian · mono · 16 kHz
← {"type":"vad","event":"speech_started","start":1.20}
← {"type":"vad","event":"speech_ended","start":1.20,"end":4.85}About the platform
VoiceKit — a cloud platform for speech: neural text-to-speech and voice cloning, timestamped transcription, real-time streaming transcription and VAD turn detection, diarization, and intelligence analysis (emotions, topics, summaries, PII redaction). It suits narration, voice assistants, call centers, interviews, and meetings.
Integration takes minutes: sign up, get an API key, and send HTTP or WebSocket requests. Synthesis returns audio in MP3, WAV, or OGG; transcription and analysis return structured JSON; streaming returns live events. The platform is optimized for Russian and detects the recording language automatically.
Voices with emotions and cloning, TTS and STT streaming, VAD detection, long-form synthesis for audiobooks, audio and video effects, webhooks for async jobs, official SDKs, and transparent pricing — from a free plan to enterprise limits.
Blog
Tutorials and deep dives on speech recognition, synthesis, diarization, and voice cloning.
From API key to timestamped transcription in Python — ready-to-run code inside.
Read article →Quality, price, and latency of three Russian TTS APIs — with a verdict.
Coming soonWho is speaking in a recording — explained simply, with API examples.
Coming soonFAQ
It's a REST platform combining speech synthesis, transcription, and audio analysis. You send a request with an API key and get a ready audio file or structured JSON.
Synthesis is optimized for Russian; transcription detects the recording language automatically. Audio is returned in MP3, WAV, and OGG.
Sign up in the dashboard — the key is created automatically. Send it in the X-Api-Key header on every request.
Yes, the Free plan includes 5,000 synthesis characters, 30 transcription minutes, and 10 analysis requests per month — no card required.
Top up your balance with crypto (USDT, BTC, ETH) and pay for any plan in your dashboard — the subscription renews automatically.
Yes. Pro and Business plans include streaming transcription and VAD over WebSocket, plus voice cloning (9 languages) via POST /v1/voices/clone.