media

Audio, image and video work: text-to-speech through an OpenAI-compatible endpoint (with a Russian fallback), audio/video transcription, image generation and editing, quick ffmpeg screenshot demo video

SKILL.md

GitHub

media — speech, image, video, and story production

Home branch in Memora: root/skill/media.

For motion graphics, animated explainers, and product promos, use the hyperframes skill first; use parts/demo-video/ here when you need a fast ffmpeg stitch of existing screenshots with TTS narration.

Choosing a part

Task Part
Russian & multilingual TTS fallback (edge-tts / local pipeline, default voice path) parts/ru-tts-fallback/
Speech synthesis via an OpenAI-compatible /v1/audio/speech endpoint parts/speech/
Audio/video transcription to text parts/transcribe/
Generate or edit images via scripts/generate.py (a local gateway or provider keys from the root .env) parts/image-generation/
Fast ffmpeg + edge-tts video stitch from HTML scenes or screenshots parts/demo-video/
Extract transcripts from YouTube videos parts/youtube-transcripts/
Extract full structured content from YouTube videos parts/youtube-full/
Short-drama stage 1: adaptation brief, cast, hook/beat table → outline.json parts/novel-outline/
Short-drama stage 2: character profiles, image/voice prompts, character sheets parts/novel-characters/
Short-drama stage 3: scene sets, consistency anchors, lighting variants parts/novel-art/
Short-drama stage 4: timed screenplay with dialogue and beat flows → script.json parts/novel-script/

Rules

  • Walk root/skill/media in Memora before starting; record working voice/image models and pipeline parameters as nodes.
  • Route TTS and image calls through the OpenAI-compatible base URL in the root .env (OPENAI_API_BASE and its key). Never pay or create an external account without the owner's approval through the host's question channel.
  • Save generated images, audio, and video inside the target project's directory, and open the deliverable for the owner with reveal_file.

Related

ai-image-prompts-skill (prompt engineering and style libraries), decks (slide presentations), design (visual composition).