Cookie Preferences
We use cookies to personalize content and analyze traffic. See our Cookie Policy.
No output yet. Submit the form to generate content.
Explore different use cases and parameter configurations
Transparent pricing with no hidden fees. Pay as you go.
| Rule & Modality | Channel | Credits | Price (USD) | Official / Reference Price | Daily Savings |
|---|---|---|---|---|---|
test otherGoogle | Regular | 999per run | $4.541 | - | - |
Complete guide to using Gemini Omni
Generate natural, expressive, and multilingual audio from text prompts using Google's omnimodal Gemini model through a single unified endpoint.
The Gemini Omni Audio API on ApiPass lets you define, name, and register a voice — timbre, tone, mood, accent, and persona — as a reusable audio profile that returns a portable audio ID. Instead of re-describing a voice every time you generate a video or animate a character, you build the voice once and then reference it across text-to-video, image-to-video, and video-edit generations with synchronized audio produced in the same forward pass. Because Gemini Omni is natively multimodal — processing text, image, audio, and video simultaneously for more cohesive, consistent, and controllable output — the voice profiles you create here stay perfectly in sync with the visuals produced by the rest of the Gemini Omni family on ApiPass.
Skip Vertex AI onboarding, billing accounts, and IAM roles. A single ApiPass API key unlocks Gemini Omni Audio instantly.
Powered by Google's flagship omnimodal architecture that natively reasons across text and audio in one model.
Built on the same generation of Gemini audio that delivers dramatically improved audio quality that feels like speaking with a person, with richer, more natural voice interactions across 30 HD voices in 24 languages.
Submit once, receive a taskId, and either poll recordInfo or use callBackUrl for push notifications — perfect for long-running generations without blocking your servers.
Speak in multiple languages and the model effortlessly switches between them without any pre-configuration; language is no longer a barrier.
Track cost, latency (costTime), and success rates across all Gemini, Veo, and Nano-Banana models in one ApiPass dashboard.
Leveraging the Gemini Omni API, generated videos accurately reflect physical, historical, and narrative rules, ensuring every scene, object, and action behaves with clear intent.
Leveraging Gemini's audio stack, which can understand and respond appropriately to users' emotional expressions for more nuanced conversations, generated output carries natural intonation and affect.
Leveraging the Gemini Omni Flash API, creators can combine text, images, video clips, and voice inputs to build structured, precise video content from real reference assets.
Every request returns a taskId; results (including resultUrls to the generated audio file) are retrievable via recordInfo or delivered directly to your callBackUrl.
Give your chatbot a natural, multilingual voice without stitching together TTS, STT, and LLM providers.
Convert scripts, articles, or long-form content into narrated audio with realistic prosody.
Auto-generate dubs and voice-overs in 20+ languages for e-learning, marketing, or gaming.
Build screen readers, real-time narration, and reading-assistance apps that sound human, not robotic.
Where the official audio surfaces are primarily optimized for real-time conversational agents, the ApiPass Gemini Omni Audio capability is optimized for asset creation — producing a reusable voice profile that plugs directly into Gemini Omni video generation.
Official access to Gemini's audio, video, and multimodal surfaces is split across different endpoints, consoles, and preview tracks. ApiPass consolidates the Omni family — audio, character, video — behind a single API key, single dashboard, and consistent JSON schema.
Replace stiff, pre-recorded IVR menus with dynamic, context-aware voice responses generated on demand.
Power SaaS products that turn blog posts, tweets, or scripts into shareable audio clips and podcast episodes.
Produce native-quality pronunciation samples and interactive dialogues across dozens of languages.
Give non-player characters unique, expressive voices generated at runtime instead of relying on pre-recorded lines.
Sign up at apipass.dev and copy your bearer token from the dashboard.
Send a POST to https://api.apipass.dev/api/v1/jobs/createTask with "model": "google/gemini-omni-audio", your input prompt, and an optional callBackUrl. You'll receive a taskId in response.
Either wait for the callback, or GET https://api.apipass.dev/api/v1/jobs/recordInfo?taskId=... to fetch the finished audio URL from the resultJson.resultUrls field once state is success.
All APIs require authentication via Bearer Token.
Authorization: Bearer
Create a reusable Gemini Omni voice configuration by selecting a preset voice and optionally adding voice characteristics and example dialogue. On success, the upstream API returns a kieAudioId that can be referenced by Gemini Omni Character or Video workflows.
The upstream KIE API accepts a direct JSON payload with the following structure. This endpoint does not use the ApiPass model/input wrapper.
1{
2 "audio_id": "string",
3 "name": "string",
4 "voice_description": "string (optional)",
5 "example_dialogue": "string (optional)"
6}audio_idRequiredstringPreset voice ID to use as the base voice. This is not the kieAudioId returned by the create endpoint.
Choose one of the 30 documented preset voice IDs. Use the returned data.kieAudioId later in Character or Video requests through the audio_ids array.
Available options:
"achernar"
nameRequiredstringName of the voice configuration. Maximum length: 210 characters.
"Gentle Narrator"
voice_descriptionOptionalstringOptional description of the voice characteristics, such as tone, style, speaking speed, emotion, or delivery. Maximum length: 20000 characters.
"A soft, clear female voice with a high pitch and a calm speaking style."
example_dialogueOptionalstringOptional example line for the voice configuration. Maximum length: 120 characters. This field does not imply that the API returns a synthesized audio file.
"Welcome to a quiet morning in the garden."
1export KIE_API_KEY='<KIE_API_KEY>'
2
3curl --request POST \
4 --url 'https://api.kie.ai/api/v1/omni/audio/create' \
5 --header "Authorization: Bearer ${KIE_API_KEY}" \
6 --header 'Content-Type: application/json' \
7 --data '{
8 "audio_id": "achernar",
9 "name": "Gentle Narrator",
10 "voice_description": "A soft, clear female voice with a high pitch and a calm speaking style.",
11 "example_dialogue": "Welcome to a quiet morning in the garden."
12 }'1{
2 "code": 0,
3 "msg": "success",
4 "data": {
5 "kieAudioId": "a8f1c2d3e4f5...",
6 "name": "gentle female voice"
7 }
8}codeUpstream status code. The official Audio example uses 0 for success; do not assume that only 200 is valid for this endpoint.
msgUpstream result message.
data.kieAudioIdThe ID of the newly created voice configuration. Pass this value to Character or Video APIs through the audio_ids field.
data.nameThe name of the created voice configuration.
