Synthesise speech

curl -X POST https://app.voho.ai/v1/speech \
  -H "Authorization: Bearer $VOHO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    "voice": "layla",
    "model": "sada-1",
    "format": "mp3"
  }' \
  --output greeting.mp3

import requests

r = requests.post(
    "https://app.voho.ai/v1/speech",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
        "voice": "layla",
        "model": "sada-1",
        "format": "mp3",
    },
)
r.raise_for_status()
open("greeting.mp3", "wb").write(r.content)

print(r.headers["X-Voho-Characters"], "chars",
      r.headers["X-Voho-Cost-Cents"], "cents")

const res = await fetch("https://app.voho.ai/v1/speech", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.VOHO_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    voice: "layla",
    model: "sada-1",
    format: "mp3",
  }),
});

if (!res.ok) throw new Error((await res.json()).error.message);
const audio = Buffer.from(await res.arrayBuffer());
await fs.writeFile("greeting.mp3", audio);

HTTP/1.1 200 OK
Content-Type: audio/mpeg
X-Voho-Engine: voho-speech/1
X-Voho-Model: sada-1
X-Voho-Voice: layla
X-Voho-Format: mp3
X-Voho-Sample-Rate: 24000
X-Voho-Characters: 34
X-Voho-Cost-Cents: 1

<binary audio>

{
  "error": {
    "code": "unknown_voice",
    "message": "Unknown voice \"laila\"."
  }
}

POST

speech

curl -X POST https://app.voho.ai/v1/speech \
  -H "Authorization: Bearer $VOHO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    "voice": "layla",
    "model": "sada-1",
    "format": "mp3"
  }' \
  --output greeting.mp3

import requests

r = requests.post(
    "https://app.voho.ai/v1/speech",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
        "voice": "layla",
        "model": "sada-1",
        "format": "mp3",
    },
)
r.raise_for_status()
open("greeting.mp3", "wb").write(r.content)

print(r.headers["X-Voho-Characters"], "chars",
      r.headers["X-Voho-Cost-Cents"], "cents")

const res = await fetch("https://app.voho.ai/v1/speech", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.VOHO_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    voice: "layla",
    model: "sada-1",
    format: "mp3",
  }),
});

if (!res.ok) throw new Error((await res.json()).error.message);
const audio = Buffer.from(await res.arrayBuffer());
await fs.writeFile("greeting.mp3", audio);

HTTP/1.1 200 OK
Content-Type: audio/mpeg
X-Voho-Engine: voho-speech/1
X-Voho-Model: sada-1
X-Voho-Voice: layla
X-Voho-Format: mp3
X-Voho-Sample-Rate: 24000
X-Voho-Characters: 34
X-Voho-Cost-Cents: 1

<binary audio>

{
  "error": {
    "code": "unknown_voice",
    "message": "Unknown voice \"laila\"."
  }
}

Returns the finished audio as raw bytes, with the details of what was produced in the response headers. For audio that starts playing before synthesis completes, use streaming instead.

Body

string

required

The text to speak. Maximum 5,000 characters.

string

required

Voice ID, for example layla. See Voices.

string

default:"sada-1"

sada-1 (flagship) or nabra-1 (fast, economical). See Models.

string

default:"mp3"

mp3, wav, opus, or mulaw. Overrides the Accept header when both are present.

integer

Output sample rate in Hz. Defaults to the format’s native rate — 24000 for mp3, wav, and opus; 8000 for mulaw.

number

default:"1.0"

Speaking rate. Below 1.0 is slower, above is faster.

number

Pitch adjustment. Only applies to nabra-1 — sada-1 ignores it.

Formats

Format	Content type	Default rate	Use it for
`mp3`	`audio/mpeg`	24000	General playback, smallest files
`wav`	`audio/wav`	24000	Uncompressed, further processing
`opus`	`audio/ogg`	24000	Low-bandwidth streaming
`mulaw`	`audio/basic`	8000	Telephony — SIP trunks, Cisco, Avaya

mulaw at 8 kHz is what telephony systems actually carry. Requesting it directly avoids a transcoding step and the quality loss that comes with it.

You can also negotiate the format with the Accept header, which is how most audio clients ask:

curl -X POST https://app.voho.ai/v1/speech \
  -H "Authorization: Bearer $VOHO_API_KEY" \
  -H "Accept: audio/basic" \
  -H "Content-Type: application/json" \
  -d '{"text":"أهلاً بك في فوهو","voice":"layla"}' \
  --output greeting.ulaw

An explicit format in the body wins over Accept.

Response

Raw audio bytes with Content-Type set for the format, plus the X-Voho-* headers describing the result.

curl -X POST https://app.voho.ai/v1/speech \
  -H "Authorization: Bearer $VOHO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    "voice": "layla",
    "model": "sada-1",
    "format": "mp3"
  }' \
  --output greeting.mp3

import requests

r = requests.post(
    "https://app.voho.ai/v1/speech",
    headers={"Authorization": f"Bearer {API_KEY}"},
    json={
        "text": "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
        "voice": "layla",
        "model": "sada-1",
        "format": "mp3",
    },
)
r.raise_for_status()
open("greeting.mp3", "wb").write(r.content)

print(r.headers["X-Voho-Characters"], "chars",
      r.headers["X-Voho-Cost-Cents"], "cents")

const res = await fetch("https://app.voho.ai/v1/speech", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${process.env.VOHO_API_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "أهلاً بك في فوهو، كيف أقدر أساعدك؟",
    voice: "layla",
    model: "sada-1",
    format: "mp3",
  }),
});

if (!res.ok) throw new Error((await res.json()).error.message);
const audio = Buffer.from(await res.arrayBuffer());
await fs.writeFile("greeting.mp3", audio);

HTTP/1.1 200 OK
Content-Type: audio/mpeg
X-Voho-Engine: voho-speech/1
X-Voho-Model: sada-1
X-Voho-Voice: layla
X-Voho-Format: mp3
X-Voho-Sample-Rate: 24000
X-Voho-Characters: 34
X-Voho-Cost-Cents: 1

<binary audio>

{
  "error": {
    "code": "unknown_voice",
    "message": "Unknown voice \"laila\"."
  }
}

Errors

Status	Code	Cause
400	`invalid_request`	Body is not valid JSON
400	`empty_text`	`text` missing or blank
400	`text_too_long`	Over 5,000 characters
400	`unknown_format`	Unrecognised `format`
401	`unauthorized`	Missing, malformed, or revoked token
402	`insufficient_credit`	Balance exhausted
404	`unknown_voice`	No such voice ID
404	`unknown_model`	No such model ID
502	`synthesis_failed`	Synthesis produced no audio
503	`unavailable`	Briefly at capacity — retry

unavailable is transient. Retry with backoff. synthesis_failed usually is not — check your request before retrying.

Authorizations

Authorization

string

header

required

API token from the Voho console. Begins with voho_sk_live_.

Body

application/json

text

string

required

Text to speak. Maximum 5,000 characters.

Maximum string length: 5000

Example:

"أهلاً بك في فوهو"

voice

string

required

Voice ID from the catalog.

Example:

"layla"

model

enum<string>

default:sada-1

Available options:

sada-1,

nabra-1

format

enum<string>

default:mp3

Overrides the Accept header when both are present.

Available options:

mp3,

wav,

opus,

mulaw

sample_rate

integer

Defaults to the format's native rate (24000, or 8000 for mulaw).

speed

number

default:1

Speaking rate. Below 1.0 is slower.

pitch

number

Pitch adjustment. Only applies to nabra-1; sada-1 ignores it.

Response

Audio bytes.

The response is of type file.

Overview

Speech

Voices

Text utilities

Speech

Voices

Text

Synthesise speech

Body

Formats

Response

Errors

Authorizations

Body

Response

​Body

​Formats

​Response

​Errors

Authorizations

Body

Response

Body

Formats

Response

Errors