skills.rest
RatnaAtika
RatnaAtikaCOMMUNITY
Explore all skills by this author →
GitHub

media-pipeline

Selects and configures providers for image generation, text-to-speech, and speech-to-text workflows.

Updated May 29, 2026
Search Tags:#image generation#text-to-speech#transcription#whisper#elevenlabs#speech-to-text#provider selection
One-click install
npx skills add https://github.com/RatnaAtika/Indonesian-ASR-11-Class-Dataset --skill media-pipeline-ratnaatika
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: media-pipeline
Source: https://github.com/RatnaAtika/Indonesian-ASR-11-Class-Dataset/tree/main/.agents/skills/media-pipeline
Command: npx skills add https://github.com/RatnaAtika/Indonesian-ASR-11-Class-Dataset --skill media-pipeline-ratnaatika

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Choosing and wiring the right provider for image generation, TTS, and speech-to-text is fragmented across dozens of vendors with different cost, latency, and privacy trade-offs. This Skill consolidates provider selection, cost control, and compliance rules into one operational guide. ## Core Features & Use Cases - Provider Selection Matrices: Decision tables for image generation (Imagen, gpt-image, SDXL, Flux), TTS (ElevenLabs, Deepgram, Coqui), and STT (Whisper, AssemblyAI, Deepgram) based on fidelity, latency, and hosting needs. - Cost & Latency Controls: Guidance on caching TTS by voice and text hash, VAD pre-processing with silero-vad, per-feature daily caps, and streaming for latency-critical paths. - Compliance Guardrails: Hard rules for voice-cloning consent, DPA requirements before sending customer audio to third parties, retention limits, and C2PA provenance tagging. - Use Case: A team building a voice-enabled support agent uses this Skill to pick Deepgram for streaming STT, ElevenLabs Flash for low-latency TTS, and configure cost caps plus PII redaction before storing transcripts. ## Quick Start Ask the agent to pick a speech-to-text provider and wire it up for transcribing support calls with diarization and PII redaction.

Frequently Asked Questions about media-pipeline

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I choose a text-to-speech provider for my app?▼

Choose based on your priority: ElevenLabs v3 for the most natural voices, Deepgram Aura or ElevenLabs Flash for low-latency streaming, and Coqui XTTS or Piper for open on-device deployment. Cache repeated phrases by voice and text hash to control cost.

What speech-to-text service should I use for transcription?▼

Whisper-v3 offers the highest accuracy across many languages, hosted by OpenAI or self-hosted. For realtime streaming use Deepgram Nova-3 or AssemblyAI Universal-2, and for on-device use whisper.cpp or Whisper GGUF models.

Should I use hosted or self-hosted models for media generation?▼

Self-host when privacy is regulated or cost outweighs quality for batch workloads. Use hosted providers with regional endpoints when latency is critical. The Skill provides decision steps for each modality.

Can I clone a voice with ElevenLabs or similar tools?▼

Voice cloning is only permitted with explicit written consent from the speaker. The Skill treats this as a hard rule and requires a documented consent flow before any cloning feature ships.

How do I reduce speech-to-text costs on long audio files?▼

Run a VAD pre-step such as silero-vad to strip silence before transcription, set per-feature daily cost caps, and sample at lower quality during development. Diarize before summarizing meetings to avoid reprocessing.

What compliance rules apply when sending customer audio to third-party APIs?▼

Never pipe customer audio to a provider whose DPA you have not signed, and never persist raw audio beyond the documented retention window. For regulated industries, prefer HIPAA-eligible options like AWS Transcribe Medical or Azure Speech.

Related Skills

Explore More Software Engineering Skills→
Lingtai-AILingtai-AI
692

preset-skill-openrouter

Documents the OpenRouter preset template's model, endpoint, and capability configuration.

Official
Basic
intrusive-memoryintrusive-memory

podcast-audio-plan

Generate podcast audio from Fountain screenplays via Claude casting and Produciesta CLI.

Official
Advanced
koenvorsterkoenvorster

gdpr-privacy

Audit GDPR/AVG compliance gaps in VorstersNV data handling and retention practices.

Community
Advanced
ferrarifankid04ferrarifankid04

system-connector

Build deterministic Node.js connectors to third-party APIs and systems.

Community
Advanced