Skip to main content

AI MTV

AI Music Television, packaged as aimtv: play songs from your local Airadio library and paint reactive diffusion video over them. AI MTV uses feedback img2img rather than conventional audio-to-video generation.

Requirements

  1. Install and run Airadio until you have at least 2 library songs, 1 ad, and 1 station-id interstitial.
  2. An NVIDIA GPU with enough free VRAM (AI MTV will check and ask you to free the GPU yourself — it never kills other processes).

Install

With pip:

pip install aimtv

Or as an isolated CLI with pipx:

pipx install aimtv
pipx install airadio

AI MTV checks for Airadio at startup. If Airadio is missing it prints both installation choices. If the library is empty, it asks you to run airadio for a bit before trying again.

Review build (local testing)

Finite playlist of 2 library songs plus 1–3 interstitials between them, rendered to a single MP4 (audio + generated video):

aimtv run --review --yes
# → ~/.local/share/aimtv/output/aimtv-review.mp4

AI MTV reconstructs and hash-verifies each song's exact Airadio lyrics and reads each interstitial's provenance record. A local Faster-Whisper model recognizes the voice on CPU, then a monotonic aligner uses those word times only as evidence for where the verified transcript is being sung. Recognition never replaces or rewrites Airadio's authoritative words.

A deterministic RAKE/IDF-style phrase ranker supplies contextual imagery without an LLM, embeddings, or a network service. Interstitial visuals use only phrases from the verified script and cannot be covered by random footage. Songs deliberately alternate between well-supported lyric anchors and unrelated dream cutaways; those cutaways are 7–16 seconds long with 3–8 seconds between them by default.

Every render writes an adjacent *.render.json manifest tying prompts and timings back to the verified lyric hash and provenance ID. Inspect a complete plan without loading the model or using the GPU:

aimtv plan --seed 42

On first run, AI MTV shows every missing model, its destination, and the approximate download size, then asks once for permission. With consent it installs SD-Turbo (~6.5 GB) and Faster-Whisper Small (~0.5 GB), showing download progress. Voice timing runs locally on CPU before the visual model loads on the GPU. Both models and all per-clip timing results are cached locally. No audio, lyrics, prompts, or inference requests are sent to a hosted service.

Models can also be installed ahead of time:

aimtv fetch-models

Doctor

aimtv doctor

Channel prompts

The editable JSON bank provides the visual style suffix and legacy random-cutaway prompts:

  • Edit this: ~/.local/share/aimtv/prompts.json (preferred; created automatically)
  • Package default: aimtv/data/prompts.json

Schema: prompts (list of channel lines), subjects (short subject tags for weighting), style_suffix (appended to every pick). Add or remove strings anytime — no code change.

Metadata

Release files for aimtv 0.4.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for aimtv 0.4.0
File Size Uploaded Provenance Actions
aimtv-0.4.0.tar.gz 42.3 kB Without provenance Details for aimtv-0.4.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for aimtv 0.4.0
File Interpreter ABI Platform Actions
aimtv-0.4.0-py3-none-any.whl Python 3 none any Details for aimtv-0.4.0-py3-none-any.whl

Total release size: 83.2 kB

Release files / aimtv-0.4.0.tar.gz

Download URL aimtv-0.4.0.tar.gz
Size 42.3 kB
Tags Source
SHA-256 checksum
How to use checksums
e04a1cee996fe111dbc491d3ab4feb872dec0dcfd0e40d0b4a9457c142834c15
BLAKE2b-256 checksum
How to use checksums
8e6751397635808bada9d076a5f0fcc5cdc733f8dcb391653859af4075366242
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.3

Release files / aimtv-0.4.0-py3-none-any.whl

Download URL aimtv-0.4.0-py3-none-any.whl
Size 40.9 kB
Tags Python 3
SHA-256 checksum
How to use checksums
97864066b7eec40d6985b61b692058357be724e7351de5d6af8f8fa0804d558a
BLAKE2b-256 checksum
How to use checksums
8a595efb765a2249a68718db1afb09a92eb5aa21837af095b732ad526d9f0b64
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.12.3

Security for aimtv 0.4.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.4.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page