-
Notifications
You must be signed in to change notification settings - Fork 14
Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Shimmer implementation -- ispras/lingvodoc-react#1116 (#1505)
* shimmer init * next * next * next step * next step * refactoring * complete * fix * checked and fixed back * xlsx * xlsx fix * fix
- Loading branch information
Showing
5 changed files
with
170 additions
and
24 deletions.
There are no files selected for viewing
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Original file line number | Diff line number | Diff line change |
|---|---|---|
| @@ -0,0 +1,114 @@ | ||
| import numpy as np | ||
| from lingvodoc.views.v2.jitter import ( | ||
| unidirectional_autowindow, | ||
| get_window_points, | ||
| pmin, pmax, | ||
| maximumPeriodFactor, | ||
| ) | ||
| from math import pi | ||
| from pdb import set_trace as A | ||
|
|
||
| maximumAmplitudeFactor = 1.6 | ||
|
|
||
| class AmplitudeTier: | ||
| def __init__(self, tmin, tmax): | ||
| self.xmin = tmin | ||
| self.xmax = tmax | ||
| self.points = [] | ||
|
|
||
| def addPoint(self, t, value): | ||
| self.points.append({ | ||
| 'number': t, | ||
| 'value': value | ||
| }) | ||
|
|
||
|
|
||
| def get_hann_windowed_rms(sound, tmid, widthLeft, widthRight): | ||
|
|
||
| if (edges := get_window_samples(sound, tmid - widthLeft, tmid + widthRight)) is None: | ||
| return None | ||
|
|
||
| imin, imax = edges | ||
|
|
||
| sumOfSquares = 0.0 | ||
| windowSumOfSquares = 0.0 | ||
| for i in range(imin, imax + 1): | ||
| t = sound['x1'] + sound['dx'] * i | ||
| width = widthLeft if t < tmid else widthRight | ||
| windowPhase = (t - tmid) / width # in [-1 .. 1] | ||
| window = 0.5 + 0.5 * np.cos(pi * windowPhase) # Hann | ||
| if sound['ny'] == 1: | ||
| windowedValue = sound['z'][0][i] * window | ||
| else: | ||
| windowedValue = 0.5 * (sound['z'][0][i] + sound['z'][1][i]) * window | ||
| sumOfSquares += windowedValue ** 2 | ||
| windowSumOfSquares += window ** 2 | ||
|
|
||
| return np.sqrt(sumOfSquares / windowSumOfSquares) | ||
|
|
||
|
|
||
| def get_window_samples(sound, tmin, tmax): | ||
| imin = np.ceil((tmin - sound['x1']) / sound['dx']) | ||
| imax = np.floor((tmax - sound['x1']) / sound['dx']) | ||
| imin = int(max(0.0, imin)) | ||
| imax = int(min(sound['nx'], imax)) | ||
| if imax - imin < 2: | ||
| return None | ||
| return imin, imax | ||
|
|
||
|
|
||
| def point_to_amplitude_period(pulse, sound, tmin, tmax): | ||
| try: | ||
| tmin, tmax = unidirectional_autowindow(pulse, tmin, tmax) | ||
| first, last = get_window_points(pulse, tmin, tmax) | ||
| if last - first < 2: | ||
| raise ValueError(f"Too few pulses between {tmin} and {tmax} seconds.") | ||
|
|
||
| amplitude = AmplitudeTier(tmin, tmax) | ||
| for i in range(first + 1, last): | ||
| p1 = pulse['t'][i] - pulse['t'][i - 1] | ||
| p2 = pulse['t'][i + 1] - pulse['t'][i] | ||
| intervalFactor = p1 / p2 if p1 > p2 else p2 / p1 | ||
| if pmin == pmax or (pmin <= p1 <= pmax and pmin <= p2 <= pmax and intervalFactor <= maximumPeriodFactor): | ||
| peak = get_hann_windowed_rms(sound, pulse['t'][i], 0.2 * p1, 0.2 * p2) | ||
| if peak is not None and peak > 0.0: | ||
| amplitude.addPoint(pulse['t'][i], peak) | ||
|
|
||
| return amplitude | ||
| except Exception as e: | ||
| raise ValueError(f"{pulse} & {sound}: not converted to AmplitudeTier.") from e | ||
|
|
||
|
|
||
| def get_shimmer_local(pulse, sound, tmin, tmax): | ||
| numberOfPeaks = 0 | ||
| numerator = 0.0 | ||
| denominator = 0.0 | ||
| try: | ||
| tmin, tmax = unidirectional_autowindow(pulse, tmin, tmax) | ||
| peaks = point_to_amplitude_period(pulse, sound, tmin, tmax) | ||
| points = peaks.points | ||
| for i in range(1, len(points)): | ||
| p = points[i]['number'] - points[i-1]['number'] | ||
| if pmin == pmax or (p >= pmin and p <= pmax): | ||
| a1 = points[i-1]['value'] | ||
| a2 = points[i]['value'] | ||
| amplitudeFactor = a1 / a2 if a1 > a2 else a2 / a1 | ||
| if amplitudeFactor <= maximumAmplitudeFactor: | ||
| numerator += abs(a1 - a2) | ||
| numberOfPeaks += 1 | ||
| if numberOfPeaks < 1: | ||
| return None | ||
| numerator /= numberOfPeaks | ||
| numberOfPeaks = 0 | ||
| for point in points[:-1]: # why -1? | ||
| denominator += point['value'] | ||
| numberOfPeaks += 1 | ||
| denominator /= numberOfPeaks | ||
| if denominator == 0.0: | ||
| return None | ||
| return numerator / denominator | ||
| except Exception as e: | ||
| if "Too few pulses between" in str(e): | ||
| return None | ||
| else: | ||
| raise Exception(f"{pulse} & {sound}: shimmer (local) not computed.") |