screencastgen

Aligner

Facade for word-level audio-text alignment.

Source: screencastgen/aligner.py

Overview

Thin facade that delegates to alignment providers registered in Alignment Registry. Provides a stable API that the rest of the codebase imports.

Functions

`align_chunk(audio_path, text, provider, language, device) -> List[WordTiming]`

Aligns audio with text to produce word-level timing data.

Parameters:

Param	Type	Default	Description
`audio_path`	`str`	—	Path to audio file
`text`	`str`	—	Transcript text
`provider`	`str`	`"whisperx"`	Alignment provider name
`language`	`str`	`"en-US"`	Language code
`device`	`str`	`"auto"`	Compute device

Returns: List[WordTiming]

Re-exported from Alignment Registry

get_alignment_provider_names()
get_default_alignment_provider()

Dependencies

Aligner
├── Alignment Registry   (provider dispatch)
├── Types                 (WordTiming)
└──▶ consumed by Pipeline Common
     └──▶ or Remote GPU Client (remote mode)