REAL-TIME AUDIO ML · RUNNING ON EDGE→ 100k requests / month

Every product deserves a soundtrack.

CassetteAI’s Music, SFX, and TTS models render a 30-second sample in under 2 seconds and a full 3-minute track in under 10 — at 44.1 kHz stereo. One API, wired into games, creator apps, and real-time pipelines.

▶ Get API key Hear it live →

FIRST-SAMPLE LATENCY

23.0ms

CASSETTE.LIVE / SESSION_0144.1 kHz·STEREO·● STREAMING

00:00

01:00

02:00

03:00

04:00

05:00

06:00

07:00

08:00

09:00

10:00

11:00

12:00

13:00

14:00

15:00

Trusted by and featured in

01 ONE SDK · THREE MODALITIES

Three engines. One tape deck.

MOD_01 / MUSIC

Adaptive music, on-device

Prompt a mood, genre, or reference. Cassette generates 1–4 minute tracks — drums, bass, harmony, lead — streamed to your app while the user plays.

44.1 kHz stereo300M paramsdeterministic seeds< 50 ms TTFA

MOD_02 / SFX

Sound effects on demand

Describe an event — "heavy door slamming in a cathedral", "retro power-up", "rainforest at dusk". Get a tight, loopable clip rolled per-frame.

100 ms – 12 sloop-safeper-frame re-rolledge inference

MOD_03 / TTS· soon

Voices with a pulse

Natural, expressive speech — zero-shot clone from a 10-second sample, emotion tags inline. Streaming-first, built for the same real-time pipelines.

zero-shot cloneemotion tokensphoneme-alignedlaunching soon

02 LIVE PLAYGROUND · NO SIGNUP

Hear it before you believe it.

PROMPT_QUEUE · 7 PRESETS · REAL CLIPS

●OUTPUT_STREAMseed: 0x539

press ▶ to generate

Smooth chill hip-hop · D Minor · 90 BPMttfa —·44.1 kHz·↓ download .wav

TIMELINES · ONE REQUEST, END-TO-ENDCOLD START

SFX

up to 30s of sound

0.000s

0.092s

Submitted

0.092s

Warm Start

0.140s

0.048s

Execution Started

0.870s

1.010s

Execution Finished

Sound effect readyTotal~1.0s

30-sec music sample

preview quality

0.000s

0.185s

Submitted

0.185s

Warm Start

0.260s

0.075s

Execution Started

1.690s

1.950s

Execution Finished

30-second music sample readyTotal< 2.0s

Full music track

up to 3 minutes

0.000s

0.215s

Submitted

0.215s

Warm Start

0.313s

0.098s

Execution Started

9.312s

9.625s

Execution Finished

Full 3-minute music track readyTotal< 10s

OUTPUT · REFERENCE QUALITY● MASTER · -3 dB

44.1kHz

Stereo · 16-bit · .wav

44,100 samples per second — the reference rate for CD-grade audio. Same fidelity your DAW expects, straight from the API.

-2.8

-3.2

-∞-24-12-6-30 dB

03 DEVELOPERS · PRODUCTION API

Two endpoints.
One API shape.

Ship CassetteAI/music-generator or cassetteai/sound-effects-generator with a single fal.subscribe() call. Works from JavaScript, Python, and plain cURL.

→Swap model ID between music, SFX, and TTS — same call shape
→44.1 kHz stereo music, up to 3-minute tracks, professional consistency
→SFX up to 30 seconds in ~1s — fast enough for games + real-time pipelines
→Pay-per-use · $0.02 / output minute for music · $0.01 / SFX generation

import { fal } from "@fal-ai/client";// Music Generator — 30-sec sample in <2s, full 3-min track in <10s.const music = await fal.subscribe("CassetteAI/music-generator", {  input: {    prompt: "Smooth chill hip-hop beat with mellow piano, deep bass. D Minor, 90 BPM.",    duration: 50,  },  logs: true,});console.log(music.data.audio_file.url);// Same SDK, SFX model — up to 30s of effect in ~1s.const sfx = await fal.subscribe("cassetteai/sound-effects-generator", {  input: { prompt: "dog barking in the rain", duration: 30 },});console.log(sfx.data.audio_file.url);