Comprehensive voice quality feature set via PraatSauce

Extracts 37 time-series voice quality measures — F0, formants, harmonic amplitudes (corrected and uncorrected), HNR in four bands, and CPP — replicating the VoiceSauce feature set via Praat's algorithms in pladdrr. Formant corrections follow Iseli & Alwan (2004); bandwidth can optionally be estimated with Hawks & Miller (1995).

Usage

trk_praatsauce(listOfFiles, beginTime = 0, endTime = 0, windowShift = 5, windowSize = 25, minF = 50, maxF = 300, formantTracking = TRUE, numFormants = 5, maxFormantHz = 5000, nominalF1 = 500, nominalF2 = 1500, nominalF3 = 2500, preEmphFrom = 50, useBandwidthFormula = FALSE, channel = 1, resample_to_16k = TRUE, windowShape = "Gaussian1", relativeWidth = 1, toFile = TRUE, explicitExt = "psa", outputDirectory = NULL, verbose = TRUE)

Arguments

listOfFiles

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

windowShift

Numeric. Frame shift in milliseconds; sets output frame rate (1000 / windowShift Hz). Default 5 ms (200 Hz).

windowSize

Numeric. Analysis window length in milliseconds. Default 25 ms.

minF

Numeric. Lower F0 bound in Hz. Default 50 Hz.

maxF

Numeric. Upper F0 bound (ceiling) in Hz. Default 300 Hz.

formantTracking

Logical. Attempt HMM formant tracking. Currently unsupported in pladdrr — a warning is issued and untracked formants are used. Default TRUE.

numFormants

Integer. Number of formants to extract (up to 5). Default 5.

maxFormantHz

Numeric. Formant ceiling in Hz. Default 5000 Hz.

nominalF1

Numeric. Reference F1 for formant tracking in Hz. Default 500 Hz.

nominalF2

Numeric. Reference F2 for formant tracking in Hz. Default 1500 Hz.

nominalF3

Numeric. Reference F3 for formant tracking in Hz. Default 2500 Hz.

preEmphFrom

Numeric. Pre-emphasis onset frequency in Hz. Default 50 Hz.

useBandwidthFormula

Logical. If TRUE, estimate bandwidths with the Hawks & Miller (1995) formula instead of Praat's Burg estimates. Default FALSE.

channel

Integer. Audio channel to use for multi-channel files. Default 1.

resample_to_16k

Logical. Resample to 16 kHz before analysis. Default TRUE.

windowShape

Character. Window shape for audio extraction. Default "Gaussian1".

relativeWidth

Numeric. Relative width of the extraction window. Default 1.0.

toFile

Logical. If TRUE, write SSFF output files and return the paths written (invisibly). If FALSE, return an AsspDataObj. Default TRUE.

explicitExt

Character. Output file extension. Default "psa".

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 37 REAL32 tracks, n_frames x 1 each. Frame rate: 1000 / windowShift Hz (default 200 Hz).

f0

Hz. Fundamental frequency (AC method; 0 = unvoiced).

F1, F2, F3

Hz. Formant frequencies.

B1, B2, B3

Hz. Formant bandwidths.

H1u, H2u, H4u

dB. Uncorrected harmonic amplitudes at 1×, 2×, 4× F0.

H2Ku, H5Ku

dB. Uncorrected spectral amplitude near 2 kHz and 5 kHz.

A1u, A2u, A3u

dB. Uncorrected amplitude at F1, F2, F3.

H1H2u, H2H4u, H1A1u, H1A2u, H1A3u, H2KH5Ku

dB. Uncorrected spectral slope differences.

H1c, H2c, H4c

dB. Formant-corrected harmonic amplitudes (Iseli & Alwan 2004).

A1c, A2c, A3c

dB. Formant-corrected amplitudes at F1, F2, F3.

H1H2c, H2H4c, H1A1c, H1A2c, H1A3c

dB. Corrected spectral slope differences.

CPP

dB. Cepstral Peak Prominence.

HNR05, HNR15, HNR25, HNR35

dB. Harmonics-to-noise ratio in bands 0–500 Hz, 0–1500 Hz, 0–2500 Hz, 0–3500 Hz.

If toFile = TRUE: character vector of output file paths, returned invisibly.

Details

Audio is optionally resampled to 16 kHz (resample_to_16k = TRUE) before analysis. Harmonic amplitudes are searched in windows of ±10% around the expected harmonic frequency. Formant corrections use the Iseli & Alwan (2004) formula applied to F1 and F2 for H1/H2/H4, and additionally F3 for A3.

References

(Iseli and Alwan 2004)

(Hawks and Miller 1995)

(Shue et al. 2011)

Examples

if (FALSE) { # \dontrun{
# Basic usage with 5ms frame shift
result <- trk_praatsauce("speech.wav", windowShift = 5, toFile = FALSE)

# Access corrected H1-H2 (breathiness measure)
plot(result$H1H2c, type = "l", main = "H1-H2 Corrected")

# Process with custom F0 range and bandwidth formula
result <- trk_praatsauce(
  "speech.wav",
  minF = 75,
  maxF = 300,
  useBandwidthFormula = TRUE,
  toFile = FALSE
)

# Batch process multiple files
trk_praatsauce(c("f1.wav", "f2.wav", "f3.wav"), toFile = TRUE)
} # }