lexigram-multimedia

Multimedia generation umbrella for Lexigram Framework — TTS, music, video, image, upscale, interpolation, and beat analysis in one DI module.

Overview

lexigram-multimedia is the orchestration layer for the Lexigram multimedia subsystem. It wires seven independent generation packages (lexigram-multimedia-tts, lexigram-multimedia-music, lexigram-multimedia-video, lexigram-multimedia-image, lexigram-multimedia-upscale, lexigram-multimedia-interpolate, lexigram-multimedia-beat) through the framework's DI container and entry-point discovery, and exposes a single MultimediaModule for application composition. Generated assets are normalized into blob storage (MediaAsset with bytes or URI), and long-running jobs are submitted to the task queue.

Full documentation: docs.lexigram.dev

Install

uv add lexigram-multimedia
# Subsystem extras (per backend, e.g. torch-based reference servers)
uv add "lexigram-multimedia-tts[elevenlabs,openai]"
uv add "lexigram-multimedia-music[ace-step-server]"
uv add "lexigram-multimedia-video[wan22-server,cogvideox-server]"

Quick Start

from lexigram import Application
from lexigram.di.module import Module, module
from lexigram.multimedia import MultimediaModule
from lexigram.contracts.multimedia import TTSProvider, TTSRequest


@module(imports=[MultimediaModule.configure()])
class AppModule(Module):
    pass


async def main() -> None:
    async with Application.boot(modules=[AppModule]) as app:
        tts = await app.container.resolve(TTSProvider)
        result = await tts.generate(TTSRequest(text="Hello from Lexigram"))
        if result.is_ok():
            asset = result.unwrap()  # MediaAsset — bytes or URI in blob storage


if __name__ == "__main__":
    import asyncio

    asyncio.run(main())

Configuration

Zero-config usage: Call MultimediaModule.configure() with no arguments to use defaults. Default backends: local-http for TTS, music, video, and image; librosa for beat analysis; rife for frame interpolation; real-esrgan for upscaling.

Option 1 — YAML file

# application.yaml
multimedia:
  tts:
    backend: "elevenlabs"
    elevenlabs_voice_id: "21m00Tcm4TlvDq8ikWAM"
  music:
    backend: "stable-audio-open"
  video:
    backend: "local-http"
  storage_path_prefix: "multimedia/"
  cache_results: true

Option 2 — Profiles + Environment Variables

export LEX_PROFILE=production
export LEX_MULTIMEDIA__TTS__BACKEND=elevenlabs
export LEX_MULTIMEDIA__STORAGE_PATH_PREFIX=multimedia/

Option 3 — Python

from lexigram.multimedia import MultimediaModule, MultimediaConfig
from lexigram.multimedia.tts.config import TTSConfig

MultimediaModule.configure(
    config=MultimediaConfig(
        tts=TTSConfig(backend="elevenlabs", elevenlabs_voice_id="21m00Tcm4TlvDq8ikWAM"),
        cache_results=True,
    )
)

Config reference

Module Factory Methods

Key Features

• One module, seven subsystems — TTS, music, video, image, upscale, interpolation, and beat analysis wired through DI
• Entry-point discovery — subsystem providers register via lexigram.multimedia.subsystems; no manual wiring
• SubsystemAccessor — single accessor to generate any media type (generate(request) -> Result[MediaAsset, MultimediaError])
• VideoAccessor / ComposeAccessor / BeatAccessor — video processing, timeline composition, and beat analysis with dedicated accessors
• Async job submission — submit() dispatches generation to the task queue with idempotency and generation events
• Blob-backed assets — results normalized into MediaAsset (bytes or URI) via lexigram-storage

Testing

from lexigram import Application
from lexigram.multimedia import MultimediaModule


async def test_boot():
    async with Application.boot(modules=[MultimediaModule.stub()]) as app:
        assert app.container is not None

Key Source Files