Sprag: Chorus

Name: Chorus
Author: Sprag

sprag/chorus

active tts 1 provider

Playground

Sprag's text-to-speech model that synthesizes 24 kHz mono speech in ten languages from built-in preset voices, voices cloned from reference audio, or voices designed from a written description.

AuthorSprag

Languages10 languages

VoicesLoading...

Statusactive

Quick start

Chorus

Create an API key

Create API key

Make a request

Pick an SDK, request mode, and language. The model and provider route are already filled in.

ElevenLabs SDK streaming TTS

Routes the ElevenLabs SDK through allmodels and streams the audio response.

import { ElevenLabsClient } from "elevenlabs";

const client = new ElevenLabsClient({
  apiKey: process.env.ALLMODELS_API_KEY,
  baseUrl: "https://api.allmodels.io/el"
});

const stream = await client.textToSpeech.stream("voice_id", {
  text: "I'm sorry, Dave. I'm afraid I can't do that.",
  modelId: "sprag/chorus",
  outputFormat: "mp3_44100_128"
}, {
  queryParams: {
    provider: {
      "only": [
        "sprag"
      ]
    }
  }
});

const reader = stream.getReader();
for (;;) {
  const { done, value } = await reader.read();
  if (done) break;
  // Write value to your response, file, or audio playback pipeline.
}

Pass provider parameters

Pass provider preferences through the compatible SDK request options when you want to pin a route to a provider.

{
  queryParams: {
    provider: {
      only: ["sprag"]
    }
  }
}

Providers

Different providers can host the same model. Choose one provider when you need a fixed backend, or let routing select among them.

ProviderSlugPer 1k charactersLatencyiModesUptime

Sprag AI 11 params

sprag

$0.0162

—

Synchronous

Sprag AI

11 documented provider parameters for this route.

Pricing

ModeLanguagePer 1k characters

Text To Speech Any $0.0162

Parameters

input body · string · required

The text to synthesize. No character cap is published; probe 2026-09-05 saw 5,000 characters succeed and 20,000 return 400 backend_error.

model body · string · required

Model name to route to; may be an alias, and is rejected if blank.

cache_salt body · string | null · optional

Prefix-cache isolation key, overwritten when ref_audio is present. Documented on https://sprag.ai/docs/tts/api.md but ABSENT from Sprag's own published SpeechRequest schema; it is accepted only because additionalProperties is true. Unprobed.

instructions body · string | null · optional

Optional voice/style guidance on a preset model; the voice description itself on chorus-design. Probe 2026-09-05 measured an effect on duration but not a directional one.

language body · string | null · optional

ISO-639 / BCP-47 tag for the input text; omit to have the model detect it. Chorus serves ten: en, zh, ja, ko, es, pt, ru, de, fr, it.

ref_audio body · string | null · optional

Inline data: URL of reference audio for voice cloning. A web URL and a data: URL without ;base64 are both 400. Belongs to the cloning model, not to chorus-voices.

ref_text body · string | null · optional

Exact transcript of the reference audio, for voice cloning.

response_format body · string | null · optional

Omit and the model's own container comes back. Omit it on a STREAMED request and you get wav.

5 enum values

Values

mp3opusflacwavpcm

speed body · number | null · optional

Playback speed multiplier; scales generated duration by roughly 1/speed. Sprag's published schema declares only exclusiveMinimum 0, but probe 2026-09-05 shows 0.2 and 5.0 rejected with backend_error while 0.25-4.0 succeed, so the documented range is enforced by the model backend rather than by request validation.

min 0.25max 4

stream_format body · string | null · optional

`audio` streams raw chunked audio; `sse` emits speech.audio.delta / speech.audio.done events; omit for one buffered response.

3 enum values

Values

sseaudio

voice body · string | null · optional

Preset speaker id. Ids are lowercase and match case-insensitively; an unknown id returns 400 with the model's valid ids in the message. On `chorus-voices` the nine are aiden, dylan, eric, ono_anna, ryan, serena, sohee, uncle_fu and vivian.

Supported languages

10 languages

English en

Chinese zh

Japanese ja

Korean ko

Spanish es

Portuguese pt

Russian ru

German de

French fr

Italian it

Voices

Features

Text to speech

Generates spoken audio from text through an OpenAI /v1/audio/speech shape, returning audio bytes rather than JSON.

text_to_speech

Streaming

Streams raw chunked audio (stream_format=audio) or base64 audio deltas over SSE (stream_format=sse), for pcm and wav output only.

streaming

Output formats

Serves mp3, flac, wav and pcm at 24 kHz mono and opus at 48 kHz mono.

output_formats

Speaking rate control

`speed` multiplies the speaking rate over the documented 0.25-4.0 range.

speed_control

Voice cloning

Clones a speaker from reference audio passed inline as ref_audio and ref_text or saved once as an organization voice.

voice_cloning

Voice design

Creates a voice from a written description through /v1/audio/voices/design previews saved as organization voices.

voice_design

Code

ElevenLabs SDK streaming TTS

Routes the ElevenLabs SDK through allmodels and streams the audio response.

import { ElevenLabsClient } from "elevenlabs";

const client = new ElevenLabsClient({
  apiKey: process.env.ALLMODELS_API_KEY,
  baseUrl: "https://api.allmodels.io/el"
});

const stream = await client.textToSpeech.stream("voice_id", {
  text: "I'm sorry, Dave. I'm afraid I can't do that.",
  modelId: "sprag/chorus",
  outputFormat: "mp3_44100_128"
}, {
  queryParams: {
    provider: {
      "only": [
        "sprag"
      ]
    }
  }
});

const reader = stream.getReader();
for (;;) {
  const { done, value } = await reader.read();
  if (done) break;
  // Write value to your response, file, or audio playback pipeline.
}