Skip to contents

Decompose speech into vocal tract, glottis, and lip radiation LP filters (GFM-IAIF)

trk_gfmiaif.Rd

Returns per-frame LP coefficient tracks for the vocal tract, glottis, and lip radiation filters using GFM-IAIF source-filter separation (Perrotin and McLoughlin 2019) . Prefer this over trk_covarep_iaif() when you need frame-level LP coefficients rather than a sample-domain glottal waveform.

Usage

trk_gfmiaif(listOfFiles, beginTime = 0, centerTime = FALSE, endTime = 0, windowShift = 10, windowSize = 32, nv = 48L, ng = 3L, d = 0.99, window = "HANN", explicitExt = "gfm", outputDirectory = NULL, toFile = TRUE, verbose = TRUE)

Arguments

listOfFiles

Character vector of audio file paths. Any format supported by av is accepted; non-native inputs are transcoded automatically.

centerTime

Logical. If TRUE, timestamps refer to window centres; if FALSE, to window starts. Default FALSE.

windowShift

Numeric. Frame shift in milliseconds; sets output frame rate (1000 / windowShift Hz). Default 10.0 ms.

windowSize

Numeric. Analysis window length in milliseconds. Must be large enough for nv + 1 samples at the audio sample rate. Default 32.0 ms.

nv

Integer. Vocal tract LPC order (1–100). Higher values capture more spectral detail but increase compute. Default 48.

ng

Integer. Glottis LPC order. The value 3 is recommended by the original authors; departing from it may degrade estimates. Default 3.

d

Numeric. Leaky integration coefficient for lip radiation (0.9–0.999). Default 0.99.

window

Character. Analysis window type: "HANN", "HAMMING", or "BLACKMAN". Default "HANN".

explicitExt

Character. Output file extension. Default "gfm".

toFile

Logical. If TRUE, write SSFF output files and return the count written. If FALSE, return an AsspDataObj (single file only). Default TRUE.

beginTime

Start time for the extracted portion in seconds. Default: NULL (beginning of signal). Note: uses beginTime/endTime (seconds) matching DSP function conventions, unlike read_audio() which uses begin/end.

endTime

The end time of the section of the sound files that should be analysed (in seconds). Use 0 for end of file.

outputDirectory

The directory where the slice file should be stored. If not defiled (NULL), the sparse slice file will placed in the same folder as the media file.

verbose

Logical. Show a progress bar (sequential path) or a progress-aware parallel apply (pbapply/pbmcapply, if installed).

Value

If toFile = FALSE: an AsspDataObj with tracks:

av_0 … av_{nv}

REAL64, vocal tract LP coefficients, n_frames × (nv + 1). Dimensionless.

ag_0 … ag_{ng}

REAL64, glottis LP coefficients, n_frames × (ng + 1). Dimensionless.

al_0, al_1

REAL64, lip radiation LP coefficients, n_frames × 2. Dimensionless.

Frame rate: 1000 / windowShift Hz (default 100 Hz). If toFile = TRUE: integer count of files written.

Details

GFM-IAIF extends classical IAIF with a wide-band glottis model that captures both glottal formant and spectral tilt. It estimates three LP filters per frame (vocal tract, glottis, lip radiation) via alternating inverse filtering steps.

ng = 3 is strongly recommended. When toFile = FALSE, only single-file input is permitted. The ng warning is issued automatically for other values.

References

Perrotin O, McLoughlin I (2019). “A Spectral Glottal Flow Model for Source-filter Separation of Speech.” In ICASSP 2019 - 2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 7160–7164. doi:10.1109/ICASSP.2019.8682625 .

Examples

if (FALSE) { # \dontrun{
# Basic usage
result <- trk_gfmiaif("speech.wav", toFile = FALSE)

# Lower vocal tract order for faster processing
result <- trk_gfmiaif("speech.wav", nv = 24, windowShift = 5.0, toFile = FALSE)

# Batch processing to SSFF files
files <- c("file1.wav", "file2.wav", "file3.mp3")
trk_gfmiaif(files, toFile = TRUE)
} # }