Music & Audio

Audio generation and processing. Generate original music tracks and sound effects from text descriptions, convert text to natural-sounding speech with multiple voice options, and transcribe audio to text. Translation and dubbing are built by composing these endpoints — see the Dubbing Pipeline.

Capability	Description	Duration/Limits
Music	AI composition (MiniMax, ElevenLabs)	10–300 seconds
SFX	Sound effects	1–30 seconds
TTS	Text-to-speech (IDA Voice, Google, Grok, Polly, GPT Audio)	max 3000 chars/request
STT	Transcription	max 240 minutes, auto language detect

Music Generation

FOTOhub generates full music tracks from a text description via two cloud providers. Requests are synchronous: the response carries a finished audio_url.

Endpoint scope

POST /v1/ai/generate/music generates music and nothing else. It reads model, prompt, duration, genre, mood, bpm, loop and instrumental — any other field in the body is ignored, including mode. There is no mastering, stem-separation or analysis mode on this endpoint, and no /music/compose endpoint. Sending mode does not switch behaviour: the request is still billed and served as a music generation.

Endpoint

POST /v1/ai/generate/music

Billing: 5 credits/min (minimax) or 12 credits/min (elevenlabs), per started minute

Parameters

Parameter	Type	Required	Default	Description
`prompt`	string	Yes	—	Detailed description of the music. Include genre, instruments, mood, energy level, and intended use case.
`model`	string	No	`"minimax"`	`"minimax"` or `"elevenlabs"`. Any other value returns `400` with the supported list.
`duration`	integer	No	`30`	Duration in seconds. Range: 10–300. Under 10 returns `400`; over 300 is clamped. Billed per started minute, so longer tracks cost more.
`genre`	string	No*	—	Genre hint: `"electronic"`, `"jazz"`, `"classical"`, `"hip-hop"`, `"ambient"`, `"rock"`, `"folk"`, `"cinematic"`. Required for `"elevenlabs"` — that provider rejects a missing genre, so the call is refused with `400` before billing.
`mood`	string	No	—	Mood hint: `"happy"`, `"melancholic"`, `"energetic"`, `"calm"`, `"dark"`, `"uplifting"`, `"mysterious"`.
`bpm`	integer	No	auto	Target tempo in BPM. Range: 60–200; outside that returns `400`. `tempo` is accepted as a deprecated alias.
`loop`	boolean	No	`false`	Generate a seamlessly loopable track.
`instrumental`	boolean	No	`true`	When true, generates instrumental-only (no vocals). Set false for vocal elements.

Pricing

Billed per started minute, per model — not on a duration ladder. A 90-second track is 2 minutes.

Model	Credits/min	Notes
MiniMax Music	5	Cloud generation, default
ElevenLabs Music	12	Higher fidelity, `genre` required

Both endpoints are synchronous: the response carries a finished audio_url. If generation fails the charge is refunded automatically.

Response

json

{
  "model": "minimax",
  "usd_charged": 0.2250,
  "billing": {
    "method": "wallet",
    "usd_charged": 0.2250,
    "usd_charged": 0,
  },
  "audio_url": "https://s1.fotohub.app/storage/v1/object/public/generations/audio/mj_xyz789.mp3",
  "duration": 30,
  "format": "mp3"
}

Rounding

Billing rounds up to the whole minute, so a 30-second and a 60-second track both cost one minute. A 61-second track costs two.

Example

PythonTypeScriptGocURL

python

import requests

response = requests.post(
    "https://apis.fotohub.app/v1/ai/generate/music",
    headers={
        "Authorization": "Bearer fh_live_your_api_key",
        "Content-Type": "application/json"
    },
    json={
        "prompt": "Upbeat electronic dance track with pulsing synths, "
                  "crisp hi-hats, deep bass drops, and euphoric buildup "
                  "sections. Suitable for a product launch video.",
        "model": "minimax",
        "duration": 60,
        "genre": "electronic",
        "mood": "energetic",
        "bpm": 128,
        "instrumental": True
    }
)

result = response.json()
print(f"Audio URL: {result['audio_url']}")
print(f"Duration: {result['duration']}s")
print(f"USD charged: ${result['usd_charged']:.4f}")

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/generate/music",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      prompt: "Upbeat electronic dance track with pulsing synths, " +
              "crisp hi-hats, deep bass drops, and euphoric buildup " +
              "sections. Suitable for a product launch video.",
      model: "minimax",
      duration: 60,
      genre: "electronic",
      mood: "energetic",
      bpm: 128,
      instrumental: true,
    }),
  }
);

const result = await response.json();
console.log("Audio URL:", result.audio_url);
console.log(`Duration: ${result.duration}s`);
console.log(`USD charged: $${result.usd_charged}`);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

func main() {
	payload := map[string]interface{}{
		"prompt":       "Upbeat electronic dance track with pulsing synths, crisp hi-hats, deep bass drops, and euphoric buildup sections.",
		"model":        "minimax",
		"duration":     60,
		"genre":        "electronic",
		"mood":         "energetic",
		"bpm":          128,
		"instrumental": true,
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/generate/music", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	fmt.Printf("Audio URL: %s\n", result["audio_url"])
	fmt.Printf("Duration: %vs\n", result["duration"])
	fmt.Printf("USD charged: %.4f\n", result["usd_charged"])
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/generate/music" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Upbeat electronic dance track with pulsing synths, crisp hi-hats, deep bass drops, and euphoric buildup sections.",
    "model": "minimax",
    "duration": 60,
    "genre": "electronic",
    "mood": "energetic",
    "bpm": 128,
    "instrumental": true
  }'

Sound Effects

Endpoint

POST /v1/ai/generate/sfx

Billing: 3 credits (fixed, regardless of duration)

Parameters

Parameter	Type	Required	Default	Description
`prompt`	string	Yes	—	Description of the sound effect. Be specific about source, environment, and characteristics.
`duration`	integer	No	`5`	Duration in seconds. Range: 1–30. Most SFX work best at 3–10 seconds.

Response

json

{
  "usd_charged": 0.1350,
  "billing": {
    "method": "wallet",
    "usd_charged": 0.1350,
    "usd_charged": 0,
  },
  "audio_url": "https://s1.fotohub.app/storage/v1/object/public/generations/sfx/sfx_r4nd0m.mp3",
  "duration": 5,
  "format": "mp3"
}

Example

PythonTypeScriptGocURL

python

import requests

response = requests.post(
    "https://apis.fotohub.app/v1/ai/generate/sfx",
    headers={
        "Authorization": "Bearer fh_live_your_api_key",
        "Content-Type": "application/json"
    },
    json={
        "prompt": "Sci-fi laser gun firing three rapid shots, "
                  "with a reverberating echo in a metallic corridor",
        "duration": 3
    }
)

result = response.json()
print(f"SFX URL: {result['audio_url']}")

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/generate/sfx",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      prompt: "Sci-fi laser gun firing three rapid shots, " +
              "with a reverberating echo in a metallic corridor",
      duration: 3,
    }),
  }
);

const result = await response.json();
console.log("SFX URL:", result.audio_url);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

func main() {
	payload := map[string]interface{}{
		"prompt":   "Sci-fi laser gun firing three rapid shots, with a reverberating echo in a metallic corridor",
		"duration": 3,
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/generate/sfx", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	fmt.Printf("SFX URL: %s\n", result["audio_url"])
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/generate/sfx" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Sci-fi laser gun firing three rapid shots, with a reverberating echo in a metallic corridor",
    "duration": 3
  }'

Text-to-Speech

Endpoint

POST /v1/ai/generate/speech

Billing: 1–2 credits per 1000 characters

Parameters

Parameter	Type	Required	Default	Description
`text`	string	Yes	—	Text to synthesize. Max 3000 characters, or 1200 for `"ida-voice"` / `"ida-voice-pro"` — see the length limits note below. Over the limit returns `400` before anything is billed.
`model`	string	No	`"google"`	TTS engine: `"google"` (fast, cost-effective), `"ida-voice-pro"` (natural, premium), `"ida-voice"` (self-hosted, voice cloning) or `"grok"` (26 multilingual voices, cheapest). `"elevenlabs"` is accepted as a legacy alias for `"ida-voice-pro"`.
`voice_id`	string	No	—	Voice preset ID. Google: `"pl-PL-Standard-A"`, `"en-US-Neural2-F"`. IDA Voice: custom voice ID from dashboard. Grok: `"eve"`, `"ara"`, `"leo"` and 23 more — see Grok voices.
`language`	string	No	`"en"`	Target language: `"pl"`, `"en"`, `"de"`, `"fr"`, `"es"`.
`speed`	number	No	`1.0`	Speech speed multiplier. Range: 0.5–2.0.
`pitch`	number	No	`0`	Pitch adjustment in semitones. Range: -10 to +10. Ignored by `"grok"`.

Pricing

Model	Credits	USD	Notes
Grok Voice	0.7	$0.0375	per 1000 characters, 26 multilingual voices
Google Cloud TTS	1	$0.0241	per 1000 characters, fast
IDA Voice Pro	2	$0.0482	per 1000 characters, natural voice, cloning

Billed per 1000 characters, not per request, and the block count is fractional above the first one — a 2500-character synthesis bills 2.5 blocks, not 3. The first block is always charged in full, so anything under 1000 characters costs one block. characters_processed in the response is the figure billed.

Length limits are transport limits

This endpoint is synchronous and the gateway cuts a proxied request at 100 seconds. "ida-voice" renders roughly 0.058s per character, so its 1200-character cap is about 70 seconds of synthesis plus headroom for the upload — a higher cap would sell a request that cannot finish. If a long call does time out the charge is refunded, but the request is still lost, so split long texts client-side.

Grok voices

All 26 Grok voices are multilingual — one voice speaks any supported language, so pick by character rather than by locale.

Female: ara, carina, celeste, eve, iris, luna, luxMale: altair, atlas, castor, cosmo, helios, helix, kepler, leo, lumen, naksh, orion, perseus, rex, rigel, sal, sirius, ursa, zagan, zenith

python

response = requests.post(
    "https://apis.fotohub.app/v1/ai/generate/speech",
    headers={"Authorization": "Bearer fh_live_your_api_key"},
    json={
        "text": "Witaj w FOTOhub!",
        "model": "grok",
        "voice_id": "eve",
        "language": "pl",
        "speed": 1.0,
    },
)
print(response.json()["audio_url"])

Grok responses report the voice used and the billed character count:

json

{
  "model": "grok",
  "usd_charged": 0.0315,
  "audio_url": "https://s1.fotohub.app/storage/v1/object/public/audio/.../grok-tts.mp3",
  "format": "mp3",
  "voice": "eve",
  "characters_processed": 61
}

Response

json

{
  "model": "google",
  "usd_charged": 0.0450,
  "billing": {
    "method": "wallet",
    "usd_charged": 0.0450,
    "usd_charged": 0,
  },
  "audio_url": "https://s1.fotohub.app/storage/v1/object/public/generations/speech/tts_k8m2n1.mp3",
  "duration": 12.4,
  "format": "mp3",
  "characters_processed": 847
}

Example

PythonTypeScriptGocURL

python

import requests

response = requests.post(
    "https://apis.fotohub.app/v1/ai/generate/speech",
    headers={
        "Authorization": "Bearer fh_live_your_api_key",
        "Content-Type": "application/json"
    },
    json={
        "text": "Witaj w FOTOhub! Nasza platforma umozliwia "
                "generowanie obrazow, wideo i muzyki za pomoca "
                "sztucznej inteligencji.",
        "model": "google",
        "voice_id": "pl-PL-Standard-B",
        "language": "pl",
        "speed": 1.0,
        "pitch": 0
    }
)

result = response.json()
print(f"Audio URL: {result['audio_url']}")
print(f"Duration: {result['duration']}s")
print(f"Characters processed: {result['characters_processed']}")

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/generate/speech",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      text: "Witaj w FOTOhub! Nasza platforma umozliwia " +
            "generowanie obrazow, wideo i muzyki za pomoca " +
            "sztucznej inteligencji.",
      model: "google",
      voice_id: "pl-PL-Standard-B",
      language: "pl",
      speed: 1.0,
      pitch: 0,
    }),
  }
);

const result = await response.json();
console.log("Audio URL:", result.audio_url);
console.log(`Duration: ${result.duration}s`);
console.log(`Characters: ${result.characters_processed}`);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

func main() {
	payload := map[string]interface{}{
		"text":     "Witaj w FOTOhub! Nasza platforma umozliwia generowanie obrazow, wideo i muzyki za pomoca sztucznej inteligencji.",
		"model":    "google",
		"voice_id": "pl-PL-Standard-B",
		"language": "pl",
		"speed":    1.0,
		"pitch":    0,
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/generate/speech", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	fmt.Printf("Audio URL: %s\n", result["audio_url"])
	fmt.Printf("Duration: %vs\n", result["duration"])
	fmt.Printf("Characters processed: %v\n", result["characters_processed"])
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/generate/speech" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Witaj w FOTOhub! Nasza platforma umozliwia generowanie obrazow, wideo i muzyki za pomoca sztucznej inteligencji.",
    "model": "google",
    "voice_id": "pl-PL-Standard-B",
    "language": "pl",
    "speed": 1.0,
    "pitch": 0
  }'

IDA Voice Cloning

With IDA Voice Pro, you can use custom cloned voices. Upload a voice sample via the FOTOhub dashboard to create a custom voice_id, then reference it in API calls. Cloned voices support all languages with natural accent preservation.

Text-to-Speech (Polly)

Budget TTS with 106 neural/generative voices across 41 languages. 10-50x cheaper than premium providers.

Endpoints

GET  /v1/ai/tts/polly/voices
POST /v1/ai/tts/polly/synthesize

Billing: 1 credit per 10,000 characters

List Voices

GET /v1/ai/tts/polly/voices?language=pl-PL

Returns available voices filtered by language code.

Response:

json

{
  "voices": [
    {
      "id": "Ola",
      "name": "Ola",
      "language": "pl-PL",
      "language_name": "Polish",
      "gender": "Female",
      "engines": ["generative", "neural"]
    }
  ],
  "count": 5
}

Synthesize Speech

Parameter	Type	Required	Default	Description
`text`	string	Yes	—	Text to synthesize. Max 10,000 characters. Supports SSML.
`voice_id`	string	No	`"Ola"`	Voice ID from the voices list.
`engine`	string	No	`"neural"`	Engine: `"neural"`, `"generative"`, or `"standard"`.
`output_format`	string	No	`"mp3"`	Format: `"mp3"`, `"ogg_vorbis"`, or `"pcm"`.
`language_code`	string	No	—	Language hint (e.g., `"pl-PL"`, `"en-US"`).

Response: Audio stream (binary) with headers:

X-Credits-Used — credits charged
X-Characters — characters processed

Example

PythonTypeScriptGocURL

python

from fotohub import FotoHub

client = FotoHub(api_key="fh_live_...")
audio = client.tts.polly(
    text="Witaj w FOTOhub!",
    voice_id="Ola",
    engine="neural"
)
# audio is bytes (mp3)
with open("output.mp3", "wb") as f:
    f.write(audio)

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/tts/polly/synthesize",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      text: "Witaj w FOTOhub! Generuje mowe za pomoca AI.",
      voice_id: "Ola",
      engine: "neural",
      output_format: "mp3",
    }),
  }
);

const audioBuffer = await response.arrayBuffer();
// Write to file (Node.js)
import { writeFileSync } from "fs";
writeFileSync("output.mp3", Buffer.from(audioBuffer));

package main

import (
	"bytes"
	"encoding/json"
	"io"
	"net/http"
	"os"
)

func main() {
	payload := map[string]interface{}{
		"text":          "Witaj w FOTOhub! Generuje mowe za pomoca AI.",
		"voice_id":      "Ola",
		"engine":        "neural",
		"output_format": "mp3",
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/tts/polly/synthesize", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	out, _ := os.Create("output.mp3")
	defer out.Close()
	io.Copy(out, resp.Body)
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/tts/polly/synthesize" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -o output.mp3 \
  -d '{
    "text": "Witaj w FOTOhub! Generuje mowe za pomoca AI.",
    "voice_id": "Ola",
    "engine": "neural",
    "output_format": "mp3"
  }'

Pricing

Engine	Cost (AWS)	Credits	Languages
Neural	$0.016/1K chars	1 per 10K chars	41
Generative	$0.030/1K chars	1 per 10K chars	41
Standard	$0.004/1K chars	1 per 10K chars	41

Polish Voices

Available Polish voices: Ola (Female, neural+generative), Ewa (Female, generative), Maja (Female, standard), Jan (Male, standard), Jacek (Male, standard).

Text-to-Speech (GPT Audio 1.5)

Premium AI voice synthesis using OpenAI's GPT Audio 1.5 model. Natural-sounding speech with voice style instructions support — describe the tone, emotion, and pace you want.

Endpoint

POST /v1/ai/generate/speech/gpt

Billing: 2 credits per request

Parameters

Parameter	Type	Required	Default	Description
`text`	string	Yes	—	Text to synthesize. Max 10,000 characters.
`voice`	string	No	`"alloy"`	Voice: `alloy`, `ash`, `ballad`, `coral`, `echo`, `fable`, `nova`, `onyx`, `sage`, `shimmer`, `verse`
`instructions`	string	No	—	Voice style instructions. E.g., "Speak calmly like an audiobook narrator" or "Energetic sports commentator tone".

Pricing

Model	Credits	USD	Notes
GPT Audio 1.5	2	$0.1072	per request, WAV HD output

Response

json

{
  "usd_charged": 0.0900,
  "audio_url": "https://s1.fotohub.app/storage/v1/object/public/audio/gpt-tts/...",
  "transcript": "The text as spoken by the model",
  "model": "gpt-audio-1.5",
  "voice": "nova",
  "format": "wav",
  "usage": {
    "input_tokens": 32,
    "output_tokens": 216,
    "total_tokens": 248
  }
}

Example

PythonTypeScriptGocURL

python

from fotohub import FotoHub

client = FotoHub(api_key="fh_live_your_key")

result = client.audio.speech(
    text="Welcome to FOTOhub! Generate images, video, and music with AI.",
    voice="nova",
    instructions="Speak warmly and naturally, like a podcast host",
)
print(f"Audio: {result.audio_url}")

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/generate/speech/gpt",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      text: "Welcome to FOTOhub! Generate images, video, and music with AI.",
      voice: "nova",
      instructions: "Speak warmly and naturally, like a podcast host",
    }),
  }
);

const result = await response.json();
console.log(`Audio: ${result.audio_url}`);
console.log(`Voice: ${result.voice}`);
console.log(`Tokens: ${result.usage.total_tokens}`);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

func main() {
	payload := map[string]interface{}{
		"text":         "Welcome to FOTOhub! Generate images, video, and music with AI.",
		"voice":        "nova",
		"instructions": "Speak warmly and naturally, like a podcast host",
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/generate/speech/gpt", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	fmt.Printf("Audio: %s\n", result["audio_url"])
	fmt.Printf("Voice: %s\n", result["voice"])
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/generate/speech/gpt" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Welcome to FOTOhub! Generate images, video, and music with AI.",
    "voice": "nova",
    "instructions": "Speak warmly and naturally, like a podcast host"
  }'

Voice Style Instructions

GPT Audio 1.5 supports freeform instructions for voice style. You can control tone, pace, emotion, and speaking style. Examples:

"Speak slowly and clearly, like a meditation guide"
"Excited and energetic, like announcing a product launch"
"Professional news anchor tone, neutral and clear"
"Whispered, intimate storytelling voice"

Speech-to-Text (Voxtral)

LLM-quality transcription using Mistral Voxtral models. Better context understanding than traditional ASR.

Endpoints

POST /v1/ai/transcribe/voxtral/transcribe
POST /v1/ai/transcribe/voxtral/summarize

Billing: per started minute of input audio, in USD

Transcribe

Parameter	Type	Required	Default	Description
`audio`	string	Yes	—	Base64-encoded audio (WAV, MP3, OGG, FLAC, WebM). Max 25MB, max 240 minutes.
`model`	string	No	`"voxtral-small"`	Model: `"voxtral-small"` (24B, quality) or `"voxtral-mini"` (3B, fast).
`language`	string	No	—	Language hint for better accuracy.
`prompt`	string	No	—	Context/instruction for the model.

Response:

json

{
  "text": "Transcribed text content here...",
  "model": "voxtral-small",
  "minutes_billed": 2,
  "cost_usd": 0.04,
  "currency": "USD",
  "tokens": { "input": 1200, "output": 150 }
}

The duration is measured from the file itself and rounded up to whole minutes — a 95-second clip bills 2 minutes. minutes_billed is the figure the charge was computed from. If the header is unreadable the charge is one minute, the smallest honest amount. A failed transcription is refunded automatically, and every request rejected with a 400 is refused before the wallet is touched.

Summarize

Same parameters as transcribe, plus:

Parameter	Type	Required	Default	Description
`format`	string	No	`"bullets"`	Summary format: `"bullets"` or `"paragraph"`.

Returns both transcription and summary in a single call. Priced as the per-minute transcription plus one fixed summarisation charge — the summary is a single completion capped at 4096 tokens, so it does not grow with the length of the audio.

Pricing

Model	Transcribe	Summarize	Quality	Speed
Voxtral Small 24B	$0.020 / minute	$0.020 / minute + $0.010	High (LLM-quality context)	~10s
Voxtral Mini 3B	$0.005 / minute	$0.005 / minute + $0.005	Good (fast)	~3s

Worked examples, Voxtral Small:

Audio length	Minutes billed	Transcribe	Summarize
40 seconds	1	$0.020	$0.030
95 seconds	2	$0.040	$0.050
60 minutes	60	$1.200	$1.210

Both legs are itemised per model in GET /v1/pricing.

Speech-to-Text (Transcription)

Endpoint

POST /v1/ai/transcribe

Billing: 0.3 credits per started minute of input audio, both models

Parameters

Parameter	Type	Required	Default	Description
`audio_url`	string	Yes	—	URL of audio file (MP3, WAV, M4A, MP4, FLAC, OGG, WebM). Max 240 minutes; longer returns `400` before billing.
`model`	string	No	`"default"`	`"default"` or `"grok"` (per-word timestamps). Both bill 0.3 credits per started minute.
`language`	string	No	`"auto"`	Source language or `"auto"` for detection.

The duration is measured from the file itself, then rounded up to whole minutes — a 90-second clip bills 2 minutes. minutes_billed in the response is the figure charged. If the header is unreadable the charge is one minute, the smallest honest amount. A failed transcription is refunded automatically.

No translate or dub mode

This endpoint transcribes in the source language. It reads only audio_url, language and model — mode, timestamps and diarize are ignored, so sending mode: "translate" or mode: "dub" returns a plain transcript and still bills for it. To translate, pass the transcript to /v1/ai/chat/completions; to dub, feed the translation to /v1/ai/generate/speech. The Dubbing Pipeline below shows both steps.

Grok transcription

model: "grok" returns a words array with a start/end offset for every word — useful for subtitles and word-accurate seeking. It bills 0.3 credits per started minute and does not do diarization or emotion analysis.

json

{
  "model": "grok",
  "usd_charged": 0.0135,
  "text": "Transkrypcja publicznego API ze znacznikami czasu.",
  "language": "pl",
  "duration": 3.7,
  "words": [
    { "text": "Transkrypcja", "start": 0.08, "end": 0.88 },
    { "text": "publicznego", "start": 0.96, "end": 1.59 },
    { "text": "API", "start": 1.73, "end": 2.11 }
  ]
}

Response

json

{
  "usd_charged": 0.1350,
  "billing": {
    "method": "wallet",
    "usd_charged": 0.1350,
    "usd_charged": 0,
  },
  "text": "Dzien dobry, chcialbym zamowic projekt graficzny dla mojej firmy...",
  "language_detected": "pl",
  "duration_minutes": 2.8,
  "segments": [
    {
      "start": 0.0,
      "end": 3.2,
      "text": "Dzien dobry, chcialbym zamowic",
      "speaker": null
    },
    {
      "start": 3.2,
      "end": 6.8,
      "text": "projekt graficzny dla mojej firmy...",
      "speaker": null
    }
  ],
  "confidence": 0.96
}

Example

PythonTypeScriptGocURL

python

import requests

response = requests.post(
    "https://apis.fotohub.app/v1/ai/transcribe",
    headers={
        "Authorization": "Bearer fh_live_your_api_key",
        "Content-Type": "application/json"
    },
    json={
        "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/meeting-recording.mp3",
        "language": "auto",
        "mode": "transcribe",
        "timestamps": True,
        "diarize": True
    }
)

result = response.json()
print(f"Detected language: {result['language_detected']}")
print(f"Full transcript: {result['text']}")

for segment in result["segments"]:
    speaker = segment.get("speaker", "Unknown")
    print(f"[{segment['start']:.1f}s - {segment['end']:.1f}s] "
          f"Speaker {speaker}: {segment['text']}")

typescript

const response = await fetch(
  "https://apis.fotohub.app/v1/ai/transcribe",
  {
    method: "POST",
    headers: {
      "Authorization": "Bearer fh_live_your_api_key",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      audio_url: "https://s1.fotohub.app/storage/v1/object/public/uploads/meeting-recording.mp3",
      language: "auto",
      mode: "transcribe",
      timestamps: true,
      diarize: true,
    }),
  }
);

const result = await response.json();
console.log(`Detected language: ${result.language_detected}`);
console.log(`Full transcript: ${result.text}`);

for (const segment of result.segments) {
  const speaker = segment.speaker ?? "Unknown";
  console.log(
    `[${segment.start.toFixed(1)}s - ${segment.end.toFixed(1)}s] ` +
    `Speaker ${speaker}: ${segment.text}`
  );
}

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

func main() {
	payload := map[string]interface{}{
		"audio_url":  "https://s1.fotohub.app/storage/v1/object/public/uploads/meeting-recording.mp3",
		"language":   "auto",
		"mode":       "transcribe",
		"timestamps": true,
		"diarize":    true,
	}
	body, _ := json.Marshal(payload)

	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/transcribe", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	fmt.Printf("Detected language: %s\n", result["language_detected"])
	fmt.Printf("Full transcript: %s\n", result["text"])
}

bash

curl -X POST "https://apis.fotohub.app/v1/ai/transcribe" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/meeting-recording.mp3",
    "language": "auto",
    "mode": "transcribe",
    "timestamps": true,
    "diarize": true
  }'

Dubbing Pipeline

A common production workflow: transcribe source audio, translate to another language, then synthesize speech in the target language. This 3-step pipeline gives you full control over each stage (edit the transcript, fix translations, choose voices).

Workflow

Transcribe — extract text from source audio with timestamps
Translate — convert transcript to the target language
Synthesize — generate speech in the target language using Premium TTS

Example

PythonTypeScriptGocURL

python

import requests

BASE = "https://apis.fotohub.app"
HEADERS = {
    "Authorization": "Bearer fh_live_your_api_key",
    "Content-Type": "application/json"
}

# Step 1: Transcribe the source audio
transcribe_resp = requests.post(
    f"{BASE}/v1/ai/transcribe",
    headers=HEADERS,
    json={
        "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
        "language": "pl",
        "mode": "transcribe",
        "timestamps": True
    }
)
transcript = transcribe_resp.json()
print(f"Transcribed ({transcript['language_detected']}): {transcript['text'][:100]}...")

# Step 2: Translate (using mode: translate for English, or use your own translation)
translate_resp = requests.post(
    f"{BASE}/v1/ai/transcribe",
    headers=HEADERS,
    json={
        "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
        "mode": "translate"
    }
)
translation = translate_resp.json()
translated_text = translation["text"]
print(f"Translated to EN: {translated_text[:100]}...")

# Step 3: Synthesize with Premium TTS (GPT Audio 1.5)
speech_resp = requests.post(
    f"{BASE}/v1/ai/generate/speech/gpt",
    headers=HEADERS,
    json={
        "text": translated_text,
        "voice": "onyx",
        "instructions": "Professional lecturer tone, clear and authoritative"
    }
)
result = speech_resp.json()
print(f"Dubbed audio: {result['audio_url']}")
total_usd_charged = (transcript["usd_charged"] +
                 translation["usd_charged"] +
                 result["usd_charged"])
print(f"Total credits: {total_usd_charged}")

typescript

const BASE = "https://apis.fotohub.app";
const HEADERS = {
  "Authorization": "Bearer fh_live_your_api_key",
  "Content-Type": "application/json",
};

// Step 1: Transcribe the source audio
const transcribeResp = await fetch(`${BASE}/v1/ai/transcribe`, {
  method: "POST",
  headers: HEADERS,
  body: JSON.stringify({
    audio_url: "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
    language: "pl",
    mode: "transcribe",
    timestamps: true,
  }),
});
const transcript = await transcribeResp.json();
console.log(`Transcribed (${transcript.language_detected}): ${transcript.text.slice(0, 100)}...`);

// Step 2: Translate to English
const translateResp = await fetch(`${BASE}/v1/ai/transcribe`, {
  method: "POST",
  headers: HEADERS,
  body: JSON.stringify({
    audio_url: "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
    mode: "translate",
  }),
});
const translation = await translateResp.json();
const translatedText = translation.text;
console.log(`Translated to EN: ${translatedText.slice(0, 100)}...`);

// Step 3: Synthesize with Premium TTS (GPT Audio 1.5)
const speechResp = await fetch(`${BASE}/v1/ai/generate/speech/gpt`, {
  method: "POST",
  headers: HEADERS,
  body: JSON.stringify({
    text: translatedText,
    voice: "onyx",
    instructions: "Professional lecturer tone, clear and authoritative",
  }),
});
const result = await speechResp.json();
console.log(`Dubbed audio: ${result.audio_url}`);
const totalUSDCharged = transcript.usd_charged + translation.usd_charged + result.usd_charged;
console.log(`Total credits: ${totalUSDCharged}`);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"net/http"
)

const base = "https://apis.fotohub.app"

func post(url string, payload map[string]interface{}) map[string]interface{} {
	body, _ := json.Marshal(payload)
	req, _ := http.NewRequest("POST", url, bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer fh_live_your_api_key")
	req.Header.Set("Content-Type", "application/json")

	resp, err := http.DefaultClient.Do(req)
	if err != nil {
		panic(err)
	}
	defer resp.Body.Close()

	var result map[string]interface{}
	json.NewDecoder(resp.Body).Decode(&result)
	return result
}

func main() {
	// Step 1: Transcribe
	transcript := post(base+"/v1/ai/transcribe", map[string]interface{}{
		"audio_url":  "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
		"language":   "pl",
		"mode":       "transcribe",
		"timestamps": true,
	})
	fmt.Printf("Transcribed (%s): %s...\n", transcript["language_detected"], transcript["text"])

	// Step 2: Translate
	translation := post(base+"/v1/ai/transcribe", map[string]interface{}{
		"audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
		"mode":      "translate",
	})
	translatedText := translation["text"].(string)
	fmt.Printf("Translated to EN: %s...\n", translatedText[:100])

	// Step 3: Synthesize with Premium TTS
	result := post(base+"/v1/ai/generate/speech/gpt", map[string]interface{}{
		"text":         translatedText,
		"voice":        "onyx",
		"instructions": "Professional lecturer tone, clear and authoritative",
	})
	fmt.Printf("Dubbed audio: %s\n", result["audio_url"])
}

bash

# Step 1: Transcribe
curl -s -X POST "https://apis.fotohub.app/v1/ai/transcribe" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
    "language": "pl",
    "mode": "transcribe",
    "timestamps": true
  }' | jq '.text' > transcript.txt

# Step 2: Translate
curl -s -X POST "https://apis.fotohub.app/v1/ai/transcribe" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d '{
    "audio_url": "https://s1.fotohub.app/storage/v1/object/public/uploads/lecture-pl.mp3",
    "mode": "translate"
  }' | jq -r '.text' > translated.txt

# Step 3: Synthesize with Premium TTS
TRANSLATED=$(cat translated.txt)
curl -X POST "https://apis.fotohub.app/v1/ai/generate/speech/gpt" \
  -H "Authorization: Bearer fh_live_your_api_key" \
  -H "Content-Type: application/json" \
  -d "{
    \"text\": \"$TRANSLATED\",
    \"voice\": \"onyx\",
    \"instructions\": \"Professional lecturer tone, clear and authoritative\"
  }"

Pipeline Costs

The manual dubbing pipeline costs approximately 5 credits per minute of source audio: 1 credit (transcription) + 2 credits (translation) + 2 credits (GPT Audio TTS). For automated single-step dubbing with voice preservation, use mode: "dub" directly (5 credits/min).

Supported Languages

All audio services support the following 23 languages for transcription, translation, TTS, and dubbing.

Code	Language	Transcription	TTS	Translation	Dubbing
`ar`	Arabic	Yes	Yes	Yes	Yes
`da`	Danish	Yes	Yes	Yes	Yes
`de`	German	Yes	Yes	Yes	Yes
`el`	Greek	Yes	Yes	Yes	Yes
`en`	English	Yes	Yes	Yes	Yes
`es`	Spanish	Yes	Yes	Yes	Yes
`fi`	Finnish	Yes	Yes	Yes	Yes
`fr`	French	Yes	Yes	Yes	Yes
`he`	Hebrew	Yes	Yes	Yes	Yes
`hi`	Hindi	Yes	Yes	Yes	Yes
`it`	Italian	Yes	Yes	Yes	Yes
`ja`	Japanese	Yes	Yes	Yes	Yes
`ko`	Korean	Yes	Yes	Yes	Yes
`ms`	Malay	Yes	Yes	Yes	Yes
`nl`	Dutch	Yes	Yes	Yes	Yes
`no`	Norwegian	Yes	Yes	Yes	Yes
`pl`	Polish	Yes	Yes	Yes	Yes
`pt`	Portuguese	Yes	Yes	Yes	Yes
`ru`	Russian	Yes	Yes	Yes	Yes
`sv`	Swedish	Yes	Yes	Yes	Yes
`sw`	Swahili	Yes	Yes	Yes	Yes
`tr`	Turkish	Yes	Yes	Yes	Yes
`zh`	Chinese	Yes	Yes	Yes	Yes

Language Auto-Detection

When language is set to "auto" (default for transcription), FOTOhub automatically detects the spoken language from the first 30 seconds of audio. The detected language code is returned in the response as language_detected.

Model Comparison

Quick reference for choosing the right model for your use case.

Category	Model	Price	Best For
Music	MiniMax	5 credits / min	Jingles, drafts, full tracks
Music	ElevenLabs Music	12 credits / min	Higher fidelity, `genre` required
SFX	ElevenLabs SFX	3 credits	Sound effects, foley — fixed per generation
TTS	Grok Voice	0.7 credits / 1K chars	Cheapest, 26 multilingual voices
TTS	Google Cloud TTS	1 credit / 1K chars	Budget TTS, fast
TTS	IDA Voice / Pro	2 credits / 1K chars	Natural voice, cloning
TTS	GPT Audio 1.5	2 credits / request	Premium quality, voice instructions
TTS	Polly (Neural)	1 credit / 10K chars	Bulk narration, 106 voices
Transcription	Default	0.3 credits / min	Diarization, emotion analysis
Transcription	Grok	0.3 credits / min	Per-word timestamps
Transcription	Voxtral Small	$0.020 / min	LLM-quality context
Transcription	Voxtral Mini	$0.005 / min	Fast transcription

Choosing a TTS Model

Grok Voice (0.7 cr/1K) — cheapest, one voice speaks any language
Google Cloud TTS (1 cr/1K) — fast, cost-effective, good for UI narration and notifications
IDA Voice Pro (2 cr/1K) — natural prosody, supports voice cloning, best for branded voices
GPT Audio 1.5 (2 cr/request) — highest quality, supports freeform style instructions, ideal for podcasts and audiobooks
Polly (1 cr/10K chars) — cheapest for bulk content, 106 voices across 41 languages

Voice Cloning — create and use cloned voices, emotional TTS, voice forensics
Shorts & Clips — audio-driven video clips with caption generation
Voice Agents — two-way spoken conversations instead of one-shot TTS: define a persona, voice and tools, then mint browser session tokens
Realtime Voice — the integration guide for those sessions: WebSocket transport, mic capture, PCM16 playback, tool calls

Pricing Summary

Service	Model	Price	Unit
Music	MiniMax	5 credits	per started minute
Music	ElevenLabs	12 credits	per started minute
Sound Effects	ElevenLabs SFX	3 credits	fixed per generation
TTS	Grok Voice	0.7 credits	per 1000 characters
TTS	Google Cloud	1 credit	per 1000 characters
TTS	IDA Voice / Pro	2 credits	per 1000 characters
TTS	GPT Audio 1.5	2 credits	per request
TTS	Polly (Neural)	1 credit	per 10,000 characters
Transcription	Default	0.3 credits	per started minute
Transcription	Grok	0.3 credits	per started minute
Transcription	Voxtral Mini	$0.005	per started minute
Transcription	Voxtral Small	$0.020	per started minute
Voxtral summarize	Mini / Small	+ $0.005 / + $0.010	once per request, on top of the per-minute rate

Character-billed endpoints charge fractional blocks above the first (2500 chars = 2.5 blocks); minute-billed endpoints round up to the whole minute, so a 95-second file bills 2 minutes. Every response reports what was charged — cost_usd on the all endpoints use usd_charged — alongside characters_processed or minutes_billed.

GET /v1/pricing returns the live figures with their units and an itemised breakdown per leg, which is the authoritative source if this table and the API ever disagree.

Error Responses

Status	Code	Description
400	`bad_request`	Invalid parameters: unsupported format, duration out of range, invalid voice_id.
402	`insufficient_funds`	Wallet balance will not cover the request. The body carries `required_usd`, `balance_usd`, `shortfall_usd` and `topup_url`.
413	`file_too_large`	Audio file exceeds 500MB. Compress or split before uploading.
422	`unprocessable_audio`	File corrupted, unsupported codec, or no detectable speech.
429	`rate_limit_exceeded`	Audio limits: 20 req/min (TTS/SFX), 10 req/min (music), 5 req/min (transcription/dubbing).

Gemini Generative TTS (Vertex AI)

Google Gemini generative speech model on Vertex AI. Features natural-language style direction ("read this cheerfully with an energetic tone"), multi-speaker dialogue scripts, and 30 prebuilt neural voices. Output format is 24 kHz mono WAV.

Billed in USD from prepaid wallet based on Google's exact token rates. Audio output is billed at 25 audio tokens per second of speech.

Endpoints

1. List Gemini Models & Token Pricing

GET /v1/ai/tts/gemini/models

Response Example

json

{
  "models": [
    {
      "id": "gemini-2.5-flash-tts",
      "name": "Gemini 2.5 Flash TTS",
      "available": true,
      "max_characters": 3000,
      "price_usd_per_1m_text_tokens": "0.15",
      "price_usd_per_1m_audio_tokens": "3.00",
      "preview": false
    },
    {
      "id": "gemini-2.5-pro-tts",
      "name": "Gemini 2.5 Pro TTS",
      "available": true,
      "max_characters": 5000,
      "price_usd_per_1m_text_tokens": "0.50",
      "price_usd_per_1m_audio_tokens": "10.00",
      "preview": false
    }
  ],
  "configured": true,
  "audio_tokens_per_second": 25
}

2. List Gemini Voices

GET /v1/ai/tts/gemini/voices

Returns the 30 prebuilt Gemini voices (Puck, Charon, Kore, Fenrir, Aoede, Leda, Zephyr, etc.) and the maximum allowed speakers for multi-speaker synthesis (max_speakers: 2).

3. Synthesize Speech with Gemini

POST /v1/ai/tts/gemini/synthesize

Request Parameters

Field	Type	Required	Default	Description
`text`	string	Yes	—	Text to synthesize. For dialogue, prefix lines with speaker names.
`model`	string	No	`"gemini-2.5-flash-tts"`	Gemini TTS model ID.
`voice`	string	No	`"Puck"`	Prebuilt voice name (used for single-speaker).
`style`	string	No	`""`	Natural language performance instructions (e.g. `"calm and whispering"`, `"energetic podcast host"`).
`speakers`	object[]	No	`null`	Array of `{speaker, voice}` mappings for dialogue (max 2 speakers).
`temperature`	float	No	`1.0`	Sampling temperature (0.0 to 2.0).

Response

Returns raw binary audio with Content-Type: audio/wav. Billing details are returned in custom headers:

X-Cost-USD: Dollar amount charged to wallet.
X-Input-Tokens: Text tokens processed.
X-Audio-Tokens: Audio tokens generated.
X-Audio-Seconds: Speech duration in seconds.
X-Model: Model used.
X-Truncated: "1" if audio hit output token ceiling, otherwise "0".

PythonTypeScriptGocURL

python

import requests

headers = {
    "Authorization": "Bearer YOUR_JWT_TOKEN",
    "Content-Type": "application/json",
}
payload = {
    "text": "Welcome to FOTOhub. What would you like to build today?",
    "voice": "Kore",
    "style": "warm, welcoming and enthusiastic",
}

resp = requests.post(
    "https://apis.fotohub.app/v1/ai/tts/gemini/synthesize",
    headers=headers,
    json=payload,
)

with open("welcome.wav", "wb") as f:
    f.write(resp.content)

print(f"Cost: ${resp.headers.get('X-Cost-USD')}, Duration: {resp.headers.get('X-Audio-Seconds')}s")

typescript

const resp = await fetch("https://apis.fotohub.app/v1/ai/tts/gemini/synthesize", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_JWT_TOKEN",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "Welcome to FOTOhub. What would you like to build today?",
    voice: "Kore",
    style: "warm, welcoming and enthusiastic",
  }),
});

const audioBlob = await resp.blob();
console.log(`Cost: $${resp.headers.get("X-Cost-USD")}`);

package main

import (
	"bytes"
	"encoding/json"
	"fmt"
	"io"
	"net/http"
	"os"
)

func main() {
	payload := map[string]string{
		"text":  "Welcome to FOTOhub. What would you like to build today?",
		"voice": "Kore",
		"style": "warm, welcoming and enthusiastic",
	}
	body, _ := json.Marshal(payload)
	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/tts/gemini/synthesize", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer YOUR_JWT_TOKEN")
	req.Header.Set("Content-Type", "application/json")

	resp, _ := http.DefaultClient.Do(req)
	defer resp.Body.Close()

	out, _ := os.Create("welcome.wav")
	defer out.Close()
	io.Copy(out, resp.Body)
	fmt.Printf("Cost: $%s\n", resp.Header.Get("X-Cost-USD"))
}

bash

curl -X POST https://apis.fotohub.app/v1/ai/tts/gemini/synthesize \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Welcome to FOTOhub. What would you like to build today?",
    "voice": "Kore",
    "style": "warm, welcoming and enthusiastic"
  }' \
  --output welcome.wav

Azure Speech Neural TTS

Microsoft Azure Speech Services with 700+ neural voices across 140 languages and dialects. Full support for SSML, emotion styles, speaking rate, pitch, and role-play voices.

Billed in USD from wallet balance per 1,000 characters submitted (tts-azure rate).

Endpoints

1. List Azure Voices

GET /v1/ai/tts/azure/voices?language=pl-PL

Query Parameters

language: Optional locale prefix (e.g. pl-PL, en-US, de-DE).

2. Synthesize with Azure

POST /v1/ai/tts/azure/synthesize

Request Parameters

Field	Type	Required	Default	Description
`text`	string	Yes	—	Plain text or `<speak>...</speak>` SSML markup (max 50,000 chars).
`voice_id`	string	No	`"pl-PL-MarekNeural"`	Azure voice identifier.
`output_format`	string	No	`"audio-24khz-96kbitrate-mono-mp3"`	Output format (`audio/mpeg` or `audio/wav`).
`style`	string	No	`null`	Expressive style (e.g. `cheerful`, `sad`, `customerservice`, `newscast`).
`rate`	string	No	`"0%"`	Speech rate adjustment (e.g. `"+10%"`, `"-15%"`).
`pitch`	string	No	`"0%"`	Pitch adjustment (e.g. `"+5Hz"`, `"-2st"`).

Response Headers

Content-Type: audio/mpeg or audio/wav
X-Cost-USD: Dollar cost
X-Characters: Character count processed

PythonTypeScriptGocURL

python

import requests

headers = {
    "Authorization": "Bearer YOUR_JWT_TOKEN",
    "Content-Type": "application/json",
}
payload = {
    "text": "Cześć! Twoje zamówienie zostało pomyślnie zrealizowane.",
    "voice_id": "pl-PL-AgnieszkaNeural",
    "style": "cheerful",
}
resp = requests.post(
    "https://apis.fotohub.app/v1/ai/tts/azure/synthesize",
    headers=headers,
    json=payload,
)
with open("order.mp3", "wb") as f:
    f.write(resp.content)

typescript

const resp = await fetch("https://apis.fotohub.app/v1/ai/tts/azure/synthesize", {
  method: "POST",
  headers: {
    Authorization: "Bearer YOUR_JWT_TOKEN",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: "Cześć! Twoje zamówienie zostało pomyślnie zrealizowane.",
    voice_id: "pl-PL-AgnieszkaNeural",
    style: "cheerful",
  }),
});
const audioBlob = await resp.blob();

package main

import (
	"bytes"
	"encoding/json"
	"io"
	"net/http"
	"os"
)

func main() {
	payload := map[string]string{
		"text":     "Cześć! Twoje zamówienie zostało pomyślnie zrealizowane.",
		"voice_id": "pl-PL-AgnieszkaNeural",
		"style":    "cheerful",
	}
	body, _ := json.Marshal(payload)
	req, _ := http.NewRequest("POST", "https://apis.fotohub.app/v1/ai/tts/azure/synthesize", bytes.NewReader(body))
	req.Header.Set("Authorization", "Bearer YOUR_JWT_TOKEN")
	req.Header.Set("Content-Type", "application/json")

	resp, _ := http.DefaultClient.Do(req)
	defer resp.Body.Close()
	out, _ := os.Create("order.mp3")
	defer out.Close()
	io.Copy(out, resp.Body)
}

bash

curl -X POST https://apis.fotohub.app/v1/ai/tts/azure/synthesize \
  -H "Authorization: Bearer YOUR_JWT_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "text": "Cześć! Twoje zamówienie zostało pomyślnie zrealizowane.",
    "voice_id": "pl-PL-AgnieszkaNeural",
    "style": "cheerful"
  }' \
  --output order.mp3

Music & Audio ​

Music Generation ​

Endpoint ​

Parameters ​

Pricing ​

Response ​

Example ​

Sound Effects ​

Endpoint ​

Parameters ​

Response ​

Example ​

Text-to-Speech ​

Endpoint ​

Parameters ​

Pricing ​

Grok voices ​

Response ​

Example ​

Text-to-Speech (Polly) ​

Endpoints ​

List Voices ​

Synthesize Speech ​

Example ​

Pricing ​

Text-to-Speech (GPT Audio 1.5) ​

Endpoint ​

Parameters ​

Pricing ​

Response ​

Example ​

Speech-to-Text (Voxtral) ​

Endpoints ​

Transcribe ​

Summarize ​

Pricing ​

Speech-to-Text (Transcription) ​

Endpoint ​

Parameters ​

Grok transcription ​

Response ​

Example ​

Dubbing Pipeline ​

Workflow ​

Example ​

Supported Languages ​

Model Comparison ​

Related APIs ​

Pricing Summary ​

Error Responses ​

Gemini Generative TTS (Vertex AI) ​

Endpoints ​

1. List Gemini Models & Token Pricing ​

Response Example ​

2. List Gemini Voices ​

3. Synthesize Speech with Gemini ​

Request Parameters ​

Response ​

Azure Speech Neural TTS ​

Endpoints ​

1. List Azure Voices ​

Query Parameters ​

2. Synthesize with Azure ​

Request Parameters ​

Response Headers ​

Music & Audio

Music Generation

Endpoint

Parameters

Pricing

Response

Example

Sound Effects

Endpoint

Parameters

Response

Example

Text-to-Speech

Endpoint

Parameters

Pricing

Grok voices

Response

Example

Text-to-Speech (Polly)

Endpoints

List Voices

Synthesize Speech

Example

Pricing

Text-to-Speech (GPT Audio 1.5)

Endpoint

Parameters

Pricing

Response

Example

Speech-to-Text (Voxtral)

Endpoints

Transcribe

Summarize

Pricing

Speech-to-Text (Transcription)

Endpoint

Parameters

Grok transcription

Response

Example

Dubbing Pipeline

Workflow

Example

Supported Languages

Model Comparison

Related APIs

Pricing Summary

Error Responses

Gemini Generative TTS (Vertex AI)

Endpoints

1. List Gemini Models & Token Pricing

Response Example

2. List Gemini Voices

3. Synthesize Speech with Gemini

Request Parameters

Response

Azure Speech Neural TTS

Endpoints

1. List Azure Voices

Query Parameters

2. Synthesize with Azure

Request Parameters

Response Headers