twat-audio 2.7.8

Audio processing plugin for TWAT

twat-audio

twat-audio is the audio-file domain package for the twat ecosystem. It preserves the existing Pedalboard resampling API and adds deterministic ffmpeg-backed helpers for common audio-file workflows.

Requirements

• Python 3.10+
• pedalboard for resampling
• ffmpeg/ffprobe for metadata, trimming, extraction, normalization, and muxing helpers

Python API

from twat_audio import AudioProcessConfig, resample_audio, normalize_audio, trim_audio

resample_audio(AudioProcessConfig("input.wav", "speech-16k.wav", target_samplerate=16000.0))
normalize_audio("input.wav", "normal.wav")
trim_audio("input.wav", "intro.wav", start=0, duration=5)

Implemented helpers:

• resample_audio: existing Pedalboard resampling behavior.
• audio_metadata: ffprobe metadata.
• normalize_audio, trim_audio, extract_audio.
• replace_audio: import/replace media audio tracks.
• simple_effect: explicit ffmpeg audio filter helper for deterministic effects.
• generate_audio: optional thin adapter to twat_genai audio providers.

Speech-specific STT, TTS, and dubbing orchestration belong in twat_speech; this package stays focused on audio files and signals.

CLI

python -m twat_audio --help
python -m twat_audio resample input.wav speech.wav --samplerate 16000
python -m twat_audio normalize input.wav normal.wav --dry-run
python -m twat_audio trim input.wav clip.wav --start 2 --duration 5 --dry-run
python -m twat_audio extract movie.mp4 voice.wav --dry-run
python -m twat_audio replace silent.mp4 voice.wav dubbed.mp4 --dry-run

Reference scripts

The audio portions of vidimportaudio, vidpredub.py, viddub-*, and vidreverb.py are represented here only where they are deterministic audio-file operations. Dubbing plans, speech provider calls, transcript handling, and subtitle work live in twat_speech.