The AI studio for film, video and music
Shoot it, score it, voice it. In one studio.
For films and trailers, YouTube and reels, ad campaigns and cinema screens. Make the shots and the stills, the voices in 37 languages, the score and the sound effects — then dub the whole thing for another market. From the app, or from your own code with an API key.
10,000 free credits. No card needed.
- speech languages, including Hindi
- 37 speech languages, including Hindi
- ready-made voices, plus your own
- 17 ready-made voices, plus your own
- free credits every month
- 10,000 free credits every month
- of credits refunded if a generation fails
- 100% of credits refunded if a generation fails
Watch with sound
Our trailer. Made entirely in Cinara.
Every shot, the score, the sound design and the voice were generated in Cinara. We cut them together.
Made by cinara.ai · 72 seconds · 16 shots, one score, 13 sound effects and one voice
Made for
The film, the channel, the campaign.
- Directors, writers and small production houses
Film making
Previsualise scenes, voice every character, score it, and lay in the sound — before anyone books a studio.
See how - Creators and channel teams
YouTube
Narration, b-roll, thumbnails, music beds and subtitles — the whole upload, made the same day.
See how - Social teams and solo creators
Reels, Shorts & TikTok
Nine-by-sixteen shots, a hook read in your voice, captions burned to the beat, a track underneath.
See how - Agencies, brand teams and founders doing their own marketing
Advertising & branding
Product stills, an ad film, a voiceover, a jingle and a brand voice that stays the same everywhere.
See how - Product teams, agencies and anyone shipping a site this week
Website rebrands
Point a script at a site, hand it an API key, and get a new look, new stills, a new voice and a launch film back.
See how - Cinemas, multiplexes and distributors
Cinema & exhibition
Pre-show reels, house announcements in every language your audience speaks, and trailers dubbed for the region.
See how
What people make with it
Six studios. One credit balance.
- 01
Voices & speech
Lifelike speech, and voices made for your work.
- Speech
- Dialogue
- Voice Changer
- Voice Design
- Voice Cloning
- Voice Library
- Singing Voice
- 02
Music & sound
Songs, sound effects and clean audio.
- Music
- Sound Effects
- Clean Audio
- 03
Transcription
Every word, who said it, and when.
- Transcribe
- Alignment
- 04
Video & dubbing
Dub, animate and bring pictures to life.
- Dubbing
- Video
- Talking Avatars
- Images
- 05
Studio audiobooks
Import a book. Hear it narrated, chapter by chapter.
- Audiobooks
- 06
Drive & sharing
Everything you make, in one place, ready to share.
- Drive
- Share links
- History
- Usage & Credits
In the app
Type it. Pick a voice. Hear it in seconds.
- Every generation shows its credits before it runs, and failed ones are refunded automatically.
- Everything you make lands in Drive, ready to play, download or share.
- History keeps what was made on the site and with each API key.
New · Singing Voice
Your cloned voice, singing.
Turn a cloned voice into a singing voice, then hear songs sung in it: your own Music songs, or a song or vocal you upload.
- 01 Start from a cloned voice. Its consent recording is the first thing the singing voice learns from.
- 02 Add about two more minutes. Read or sing. Each recording starts with its own consent statement, read by the same person.
- 03 Hear any song in it. Keep the key, or go an octave lower or higher. Add pitch correction to your own vocal take.
For developers
Your speech code already works.
Switch from ElevenLabs by changing the base URL and key. Text to speech, streaming, speech to text, forced alignment and sound effects accept the same request formats.
Works with
- Claude
- ChatGPT
- VS Code
Make speech, music, sound effects, images and video from Claude and VS Code over MCP, or from a Custom GPT in ChatGPT.
curl -X POST "https://api.cinara.ai/v1/text-to-speech/mentor" \
-H "xi-api-key: $CINARA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"text": "Hello from Cinara."}' --output hello.mp3Safety & consent
A voice belongs to its speaker.
- 01
Consent before cloning
The speaker reads a one-time consent statement on the recording. No statement heard, no voice.
- 02
One speaker, checked
Cinara checks the code words, the statement and that exactly one person is speaking.
- 03
Well-known voices blocked
Names of well-known people are refused for cloned voices, designed voices, images, video and avatars.
- 04
A record for every voice
Each cloned voice keeps its consent record: the statement, what was heard, and who made it, when.
Music refuses well-known artists in styles and lyric ideas, and every account is tied to a verified phone number. Something wrong? Report it.
Make the first thing today.
Start with 10,000 free credits. No card needed.
Sign-up needs a confirmed email and a verified phone number.