Speaker profiles
Create voice profiles from a short sample — 3–10 seconds of clean speech.
Voice ID
Identify the speaker by voice: verify identity, match a speaker against profiles and defend against spoofing.
Profile: Anna · voice_8f3c
Create voice profiles from a short sample — 3–10 seconds of clean speech.
Match the speaker against profiles using cosine similarity of embeddings.
An estimate of the probability that the voice is synthesized or recorded.
3–10 seconds of clean speech is enough for a voiceprint.
Match audio to a profile (1:1) and get similarity plus a verdict.
Find the closest profile among yours (1:N) in one request.
API
Create a voice profile and verify a speaker — three SDK calls.
# Регистрация голосового профиля
curl -X POST https://ttsapi.ru/v1/voice-id/enroll \
-H "X-Api-Key: rtt_…" \
-F "audio=@sample.wav"
# Верификация (1:1)
curl -X POST https://ttsapi.ru/v1/voice-id/verify \
-H "X-Api-Key: rtt_…" \
-F "audio=@voice.wav" \
-F "profile_id=voice_…"from voicekit import VoiceKitClient
client = VoiceKitClient(api_key="rtt_…")
profile = client.enroll_voice("sample.wav", name="Анна")
check = client.verify_voice("voice.wav", profile["profile_id"])
print(check["similarity"]) # 0.93import { VoiceKitClient } from "voicekit-client";
const client = new VoiceKitClient({ apiKey: "rtt_…" });
const profile = await client.enrollVoice("sample.wav", "Анна");
const check = await client.verifyVoice("voice.wav", profile.profile_id);
console.log(check.similarity); // 0.93using VoiceKit;
var client = new VoiceKitClient("rtt_…");
var profile = await client.EnrollVoiceAsync("sample.wav", name: "Анна");
var check = await client.VerifyVoiceAsync("voice.wav", profile!["profile_id"]!.GetValue());
Console.WriteLine(check?["similarity"]); FAQ
3–10 seconds of clean speech is enough to create a voice profile.
It estimates the probability the voice is synthesized or recorded; Russian-language quality depends on the model.
Up to 10 on Pro and up to 100 on Business — the limit depends on the plan.
It estimates the probability the audio is synthesized or replayed from a recording and returns it as ai_probability.
Verification and identification are available on Pro and Business plans.