Do I need extra software to use this skill?

Yes, this skill requires FFmpeg to be installed on your system as it utilizes the FFmpeg command-line tool for all audio processing stages.

Why does it convert audio to 16kHz WAV?

Most state-of-the-art transcription models, such as OpenAI's Whisper, are optimized for 16kHz mono audio. This format ensures the best balance between audio quality and processing efficiency.

Does it support stereo audio files?

Yes, the pipeline includes an automatic mono downmix stage that combines stereo channels into a single mono channel, as required by most speech recognition engines.

What loudness standards does this skill use?

It uses the EBU R128 standard with a target integrated loudness of -16 LUFS and a true peak of -1.5 dBTP, which are optimized levels for speech transcription.

Audio Normalize for Transcription

Name: Audio Normalize for Transcription
Author: rayshan

byrayshan

0•

データサイエンスとML

Preprocesses and normalizes audio files using FFmpeg to optimize them for high-accuracy speech-to-text transcription.

This skill provides a specialized FFmpeg-driven pipeline designed to prepare raw audio for machine learning transcription models. It automates a multi-stage process including mono downmixing, low-frequency rumble removal, noise reduction, and precise two-pass EBU R128 loudness normalization. By resampling the output to a 16kHz WAV format, it ensures maximum compatibility and accuracy with leading speech recognition systems like OpenAI's Whisper, effectively cleaning up background noise and inconsistent volume levels in a single automated workflow.

主な機能

01Automated 80Hz high-pass filtering to remove low-end rumble

020 GitHub stars

03Comprehensive before/after audio processing summary generation

04Mono downmixing and 16kHz resampling for ML model compatibility

05Stationary noise reduction using FFT-based denoiser

06Two-pass EBU R128 loudness normalization to -16 LUFS

ユースケース

01Preparing podcast recordings for automated transcription services

02Cleaning up meeting audio with inconsistent volumes for speech-to-text

03Standardizing voice-over files for AI model training and analysis

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add rayshan/agent-plugins audio-normalize

For use in Claude.ai and ChatGPT

Download Skill