Skip to contents

Ltas

Source: R/ltas-wrapper.R
Ltas.Rd

Represents a long-term average spectrum (LTAS): the frequency spectrum of a sound averaged over its whole duration.

Arguments

.xptr

Not for direct use. External pointer to the underlying C++ object; set internally when a method returns a new Ltas.

Value

An Ltas object with methods for long-term average spectrum analysis.

Details

An Ltas collapses a recording into a single spectral shape: how much energy sits in each frequency band once the moment-to-moment detail is averaged out. It's the standard tool for describing overall spectral shape (tilt, resonance pattern, noise floor), useful for comparing voice quality across speakers or checking a recording's frequency balance.

Usage


ltas <- sound$to_ltas(bandwidth = 100)
ltas <- sound$to_ltas_pitch_corrected(pitch_floor = 75, pitch_ceiling = 600)

Units and interpolation

Power values default to decibels (unit = "dB"); pass "energy" or "sones" for Praat's other two averaging units. Methods that locate a peak or trough take an interpolation of "none", "parabolic" (the default), "cubic", "sinc70", or "sinc700" - parabolic and cubic refine the peak's position between bins instead of snapping to the nearest one. get_value_at_frequency() takes a plain interpolate flag (TRUE/FALSE) rather than a named method.

Frequency domain

  • get_lowest_frequency(), get_highest_frequency() - the frequency range, in Hz

  • get_frequency_range() - the width of that range, in Hz

  • get_number_of_bins() - number of frequency bins

  • get_bin_width() - width of each bin, in Hz

  • get_frequency_from_bin(bin) - centre frequency of a bin

  • get_bin_from_frequency(frequency) - the bin nearest a frequency

Query values

  • get_value_at_frequency(frequency, unit, interpolate) - power at a given frequency

  • get_value_in_bin(bin) - power in one bin, read directly with no interpolation

  • get_minimum(fmin, fmax, unit, interpolation), get_maximum(fmin, fmax, unit, interpolation) - lowest/highest power in a frequency range

  • get_frequency_of_minimum(fmin, fmax, interpolation), get_frequency_of_maximum(fmin, fmax, interpolation) - where those extremes occur

  • get_mean(fmin, fmax, unit) - average power across a frequency range

  • get_standard_deviation(fmin, fmax, unit) - how much the power varies across a frequency range

  • get_slope(f1min, f1max, f2min, f2max, unit) - the difference in average power between two bands, e.g. spectral tilt (high-frequency band minus low-frequency band)

  • get_local_peak_height(environment_min, environment_max, peak_min, peak_max, unit) - how far a peak rises above its surrounding band, e.g. harmonic-to-noise prominence

Batch queries

Faster than calling the single-value methods in a loop, since each call crosses into C++ once for the whole set of frequencies or ranges.

  • get_values_at_frequencies(frequencies, interpolation) - values at many frequencies at once

  • get_means_batch(fmins, fmaxs, unit) - means for many frequency ranges at once

  • get_peaks_batch(fmins, fmaxs, interpolation), get_minima_batch(fmins, fmaxs, interpolation) - peaks or troughs for many ranges at once, returned as a data frame

Trend and peaks

  • compute_trend_line(fmin, fmax) - the fitted trend line, as its own Ltas

  • subtract_trend_line(fmin, fmax) - the spectrum with that trend removed, e.g. to flatten spectral tilt before comparing bands

  • report_spectral_trend(fmin, fmax, frequency_scale, fit_method) - slope, intercept, and fit quality (R-squared, residual error) as a structured result; print it directly or read $fitted_values for the per-point predictions

  • get_spectral_slope(fmin, fmax, frequency_scale, fit_method) - just the slope from the above, when that's all you need

  • to_spectrum_tier_peaks() - the local maxima across the whole spectrum, as a SpectrumTier

Export

  • as_data_frame() - frequency and power as a data frame

  • as_matrix() - frequency and power as a plain matrix

  • to_matrix() - the Ltas converted to a full Matrix object

  • save(path) - write to a Praat text file

See also

Sound, Spectrum, Spectrogram, SpectrumTier

Examples

sound <- Sound$create_tone(frequency = 150, duration = 0.5)
ltas <- sound$to_ltas(bandwidth = 100)
ltas$get_mean()
#> [1] -29.29521
ltas$get_number_of_bins()
#> [1] 221