Module timing 

Phoneme timing extraction from ONNX model duration output.

VITS models optionally output a durations tensor [1, phoneme_length] containing the number of frames (hop_length-sized) each phoneme occupies. This module converts frame counts to millisecond timestamps.

Structs

PhonemeTimingInfo

Timing information for a single phoneme

TimingResult

Complete timing result for a synthesized utterance

Constants

DEFAULT_HOP_LENGTH

Default hop length for VITS models

Functions

durations_to_timing

Convert duration tensor output to timing information.