-
Notifications
You must be signed in to change notification settings - Fork 0
/
2) ProPerlyAsked preparation.Rmd
379 lines (299 loc) · 14.9 KB
/
2) ProPerlyAsked preparation.Rmd
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
# ProPer preparation (II): Praat-to-R
Collect data from different objects into an R dataframe.
```{r clean_start, warning=FALSE}
rm(list = ls())
## Load required libraries
require(rPraat)
require(tidyverse)
require(zoo)
require(readxl)
require(seewave)
```
Get the timing skeleton
```{r fullTime}
files_intensity <- list.files(path="praat_data/intensity_tiers/", pattern="*.IntensityTier",full.names=T)
##### Full-time
# (get the full time table of each audio file based on its intensity tier)
fullTime_df <- plyr::ldply(files_intensity, function(f){
filename <- str_match(f,".*/([^/.]*)\\.[^/]*$")
file <- filename[,2]
it <- it.read(f)
time <- seq(it[["tmin"]], it[["tmax"]], 0.001) * 1000
data.frame(file, t=as.integer(as.character(time)))#, speaker, variable)
})
files <- fullTime_df$file
files <- files[!duplicated(files)==TRUE]
```
# Add intensityTiers, PitchTiers and pitchObjects
```{r tiers_data, warning=FALSE}
##########--- Read and collect Praat data into R tables ---##########
##### Intensity
# (get intensity data)
intensity_df <- plyr::ldply(files_intensity, function(f){
filename <- str_match(f,".*/([^/.]*)\\.[^/]*$")
file <- filename[,2]
intensitier <- it.read(f)
time = round(intensitier$t,3)*1000
data.frame(file, t=as.integer(as.character(time)), intensity = round(intensitier$i,4))
})
fullTime_df <- left_join(fullTime_df, intensity_df, by = c("file", "t"))
##### F0: Pitch Tier
##
files_pitchTier <- list.files(path="praat_data/pitch_tiers/", pattern="*.PitchTier",full.names=T)
# (get the smooth F0 curve)
f0_smooth_df <- plyr::ldply(files_pitchTier, function(f){
filename <- str_match(f,".*/([^/.]*)\\.[^/]*$")
file <- filename[,2]
pt <- pt.read(f)
time = round(pt[["t"]],3)*1000
f0 = pt[["f"]]
data.frame(file, t=as.integer(as.character(time)), f0_smooth=round(f0,2))
})
fullTime_df <- left_join(fullTime_df, f0_smooth_df, by = c("file", "t"))
#fullTimeF0_df <- left_join(fullTime_df, f0_smooth_df, by = c("file", "t"))
# fixTimed_df <- left_join(fixTimed_df, intensity_df, by = c("file", "t"))
# fixTimed_df <- left_join(fixTimed_df, f0_smooth_df, by = c("file", "t"))
```
# Read TextGrids: 'syll' tier
```{r pre_prepare_TextGrids_Syllable, warning=FALSE}
##### TextGrid
# TextGrids are optional! They are useful for exposition and to improve the automatic detection. By default, TextGrids are expected with at least one interval tier demarcating syllabic boundaries
#
files_textGrid <- list.files(path="praat_data/textgrids/", pattern="*.TextGrid",full.names=T)
# The following chunk takes syllabic intervals and labels from the "Syllable" tier
textGridSyll_df <- plyr::ldply(files_textGrid, function(f){
filename <- str_match(f,".*/([^/.]*)\\.[^/]*$")
file <- filename[,2]
tg <- tg.read(f, encoding = "auto")
syll_tier <- data.frame(tg$syll)
t1 <- ifelse(syll_tier$label=="", NA, round(syll_tier$t1,3)*1000)
t2 <- ifelse(syll_tier$label=="", NA, round(syll_tier$t2,3)*1000)
t_mid <- round((t1+t2)/2)
syll_label <- syll_tier$label
data.frame(file, t=as.integer(as.character(t1)), syll_start=as.integer(as.character(t1)), syll_mid=as.integer(as.character(t_mid)), syll_end=as.integer(as.character(t2)), syll_bounds=as.integer(as.character(t1)), syll_label)
})
textGridSyll_df <- mutate(
group_by(textGridSyll_df,file),
syll_bounds = ifelse(
(is.na(syll_bounds) & !is.na(lag(syll_end,1))),
lag(syll_end,1),
syll_bounds),
t = syll_bounds
)
#
textGridSyll_df <- dplyr::filter(textGridSyll_df, !is.na(t))
textGridSyll_df <- dplyr::filter(textGridSyll_df, syll_bounds!=0)
textGridSyll_df <- mutate(
group_by(textGridSyll_df, file),
syll_start = ifelse(syll_label=="" | syll_label==" ", NA, syll_start),
syll_mid = ifelse(syll_label=="" | syll_label==" ", NA, syll_mid),
syll_end = ifelse(syll_label=="" | syll_label==" ", NA, syll_end)
)
fullTime_df <- left_join(fullTime_df, textGridSyll_df, by = c("file", "t"))
# fullTime_df <- mutate(
# group_by(fullTime_df, file),
# stay = ifelse(t < min(syll_start, na.rm = T) | t > max(syll_end, na.rm = T), "GO", "stay")
# )
## split to fullTimed
fullTimed_df <- mutate(
group_by(fullTime_df, file),
syll_start = na.locf(syll_start, na.rm = F),
syll_mid = na.locf(syll_mid, na.rm = F),
syll_end = na.locf(syll_end, na.rm = F),
syll_bounds = na.locf(syll_bounds, na.rm = F),
syll_label = na.locf(syll_label, na.rm = F)
)
## split to fixTimed ... probably not useful anymore
# fixTimed_df <- dplyr::filter(fullTimed_df, stay=="stay")
# fixTimed_df <- droplevels(subset(fixTimed_df, select = -stay))
```
Load file to save time:
```{r PitchObject_load}
pitchObject_df <- read.csv("data_tables/pitchObject_df.csv")
```
... or run the (long) process to create 'pitchObject_df'
```{r PitchObject_run, warning=FALSE}
# ##### Pitch object
# ##
# files_pitchObject <- list.files(path="praat_data/pitch_objects/", pattern="*.Pitch",full.names=T)
# ##
# # (get the Strength, i.e. the *similarity index* or *periodic fraction* from Praat's autocorrelation. Also, get the raw pre-smoothing F0)
# pitchObject_df <- plyr::ldply(files_pitchObject, function(f){
# filename <- str_match(f,".*/([^/.]*)\\.[^/]*$")
# file <- filename[,2]
# pitch_object <- pitch.read(f)
# time <- round(pitch_object$t,3)*1000
# pitch_ceiling <- 1000 #fixed to periods up to 1000Hz
# strengthArray <- apply(as.data.frame(pitch.toArray(pitch_object)[["strengthArray"]]), 2, function(x) ifelse(x==0,NA,x))
# freqArray <- apply(as.data.frame(pitch.toArray(pitch_object)[["frequencyArray"]]), 2, function(x) ifelse(x==0,NA,x))
# zero_one_freqs <- apply(freqArray, 2, function(x) ifelse(x>pitch_ceiling, 0, 1))
# strength_limited <- strengthArray
# strength_limited[, -1] <- mapply(`*`, strengthArray[, -1], zero_one_freqs[, -1])
# ###### rowmax = highest strength value within the frequency range (up to 'pitch_ceiling')
# strength_rowmax <- apply(strength_limited, 2, max, na.rm=T)
# strength_rowmax[is.infinite(strength_rowmax)] <- 0
# data.frame(file, t=as.integer(as.character(time)), periodicStrength=round(strength_rowmax,7))
# })
#
# write.csv(pitchObject_df, "data_tables/pitchObject_df.csv", row.names=FALSE)
# ##
# ###
fullTimed_df <- left_join(fullTimed_df, pitchObject_df, by = c("file", "t"))
```
<!-- # Add F0 from the CC tool -->
<!-- ```{r F0_CC} -->
<!-- F0_Constant_df <- read.csv("data_tables/data_long_F0_CC.csv") -->
<!-- F0_Constant_df <- mutate( -->
<!-- group_by(F0_Constant_df, filename), -->
<!-- t = round((start + step)*1000), -->
<!-- f0_CC = round(f0, 2) -->
<!-- ) -->
<!-- F0_Constant_df <- droplevels(subset(F0_Constant_df, select = -c(start, end, step, stepnumber, f0, jumpkilleffect, err))) -->
<!-- colnames(F0_Constant_df) <- c("file","interval_label","t","f0_CC") -->
<!-- ##### -->
<!-- fullTimeF0_CC_df <- left_join(fullTime_df, F0_Constant_df, by = c("file", "t")) -->
<!-- fullTimeF0_CC_df <- mutate( -->
<!-- group_by(fullTimeF0_CC_df, file), -->
<!-- ## f0_smooth_stretch ('f0_smootch') -->
<!-- f0_CCsmootch = ifelse( -->
<!-- (is.na(f0_CC) & t < min(t[which(!is.na(f0_CC))])), -->
<!-- f0_CC[min(which(!is.na(f0_CC)))], ifelse( -->
<!-- (is.na(f0_CC) & t >= max(t[which(!is.na(f0_CC))])), -->
<!-- f0_CC[max(which(!is.na(f0_CC)))], f0_CC)), -->
<!-- ## f0_smooth_stretch_interp ('f0_smootchInterp') -->
<!-- f0_CCsmootchInterpLin = round(na.approx(f0_CCsmootch),2), -->
<!-- f0_CCsmootchInterpSplin = round(na.spline(f0_CCsmootch),2), -->
<!-- ## f0 post smooth_stretch_interp_smooth ('f0Post') -->
<!-- # f0_CCPost = round(bwfilter(wave = f0_CCsmootchInterpLin, f = 1000, to = 12, n = 1, output = "Sample"), 2) -->
<!-- ) -->
<!-- # fullTimeF0_CC_df <- droplevels(subset(fullTimeF0_CC_df, select = -f0_CCsmootch)) -->
<!-- fixTimedF0_df <- left_join(fixTimed_df, fullTimeF0_CC_df, by = c("file", "t")) -->
<!-- ``` -->
new:
```{r tiers_data2}
# fixTimedF0_df <- left_join(fixTimedF0_df, fullTimeF0_df, by = c("file", "t"))
fullTimed_df <- mutate(
# group_by(fixTimedF0_df, file),
group_by(fullTimed_df, file),
## f0_smooth_stretch ('f0_smootch')
f0_smootch = ifelse(
(is.na(f0_smooth) & t < min(t[which(!is.na(f0_smooth))])),
f0_smooth[min(which(!is.na(f0_smooth)))], ifelse(
(is.na(f0_smooth) & t >= max(t[which(!is.na(f0_smooth))])),
f0_smooth[max(which(!is.na(f0_smooth)))], f0_smooth)),
## f0_smooth_stretch_interp ('f0_smootchInterp')
f0_smootchInterpLin = round(na.approx(f0_smootch, na.rm=F),2),
f0_smootchInterpSplin = round(na.spline(f0_smootch, na.rm=F),2)
)
```
previously:
```{r tiers_data2, warning=FALSE}
# fullTimeF0_df <- mutate(
# group_by(fullTimeF0_df, file),
# ## f0_smooth_stretch ('f0_smootch')
# f0_smootch = ifelse(
# (is.na(f0_smooth) & t < min(t[which(!is.na(f0_smooth))])),
# f0_smooth[min(which(!is.na(f0_smooth)))], ifelse(
# (is.na(f0_smooth) & t >= max(t[which(!is.na(f0_smooth))])),
# f0_smooth[max(which(!is.na(f0_smooth)))], f0_smooth)),
# ## f0_smooth_stretch_interp ('f0_smootchInterp')
# f0_smootchInterpLin = na.approx(f0_smootch, na.rm=F),
# f0_smootchInterpSplin = na.spline(f0_smootch, na.rm=F)
# )
#
# # fullTimeF0_df <- droplevels(subset(fullTimeF0_df, select = -f0_smootch))
#
# fixTimedF0_df <- left_join(fixTimedF0_df, fullTimeF0_df, by = c("file", "t"))
```
# Add data from the big harvesting table
```{r TVNA_table}
####
# rm(list = c("bigTable","midTable","smallTable"))
bigTable <- read_xlsx("data_tables/Edited_harvest(mk_III).xlsx")
bigTable <- mutate(
group_by(bigTable),
file = paste0("cut-",video)
)
midTable <- droplevels(subset(bigTable, select = -c(...1, i, start, end, prelink, textlink, video, link, command, getImgCommand, thumbnail, X, href, text, startT, endT, ...18)))
colnames(midTable) <- c("startApprox","Validity","Auxiliary","Comment","Speaker_sex_age_name_role","Actual_Time","file")
midTable <- mutate(
group_by(midTable, file),
phraseTime = ifelse(!is.na(Actual_Time), Actual_Time, as.character(startApprox)),
phraseTime = as.integer(ifelse(phraseTime=="10 & 15", "10", phraseTime))
)
midTable <- droplevels(subset(midTable, select = -c(startApprox, Actual_Time)))
smallTable <- dplyr::filter(midTable, Validity=="yes")
smallTable <- mutate(
group_by(smallTable),
sex = str_split_i(Speaker_sex_age_name_role, ", *",1),
age = str_split_i(Speaker_sex_age_name_role, ", *",2),
speaker = str_split_i(Speaker_sex_age_name_role, ", *",3),
role = str_split_i(Speaker_sex_age_name_role, ", *",4)
## age = str_match(Speaker_sex_age_name_role, "[:digit:]{2}")[,1]#,
)
smallTable <- mutate(
group_by(smallTable, file),
speaker = ifelse(age=="30. Allie Beth Stuckey", "Allie Beth Stuckey", speaker),
speaker = ifelse(speaker=="X" | speaker=="XXXX01", NA, speaker),
age = ifelse(age=="30. Allie Beth Stuckey", "30", age),
age = ifelse(age=="X", NA, age),
sex = ifelse(sex=="m ", "m", sex),
role = ifelse(role=="X", NA, role),
)
smallTable$sex <- as.factor(smallTable$sex)
smallTable$age <- as.integer(smallTable$age)
smallTable$speaker <- as.factor(smallTable$speaker)
smallTable$role <- as.factor(smallTable$role)
smallTable <- droplevels(subset(smallTable, select = -Speaker_sex_age_name_role))
########
# > levels(smallTable$sex)
# [1] "f" "m"
# > levels(as.factor(smallTable$age))
# [1] "10" "20" "30" "40" "50" "60" "70" "80"
# > levels(smallTable$speaker)
# [1] "Aaron Jon Hyland" "Ahgha ?" "Allie Beth Stuckey" "Ammon Bundy"
# [5] "Andrea Tantaros" "Andy Levy" "Ann Lazarus" "Bill O’Reilly"
# [9] "Bob Beckel" "Chris Hayes" "Chris Hwang" "Chris Matthews"
# [13] "Cyndi Tiedt" "Darryl Honda" "Dennis ?" "Don Lemon"
# [17] "Donald Trump" "Donny Deutsch" "Dr. Phil" "Dylan Ratigan"
# [21] "Ellen DeGeneres" "Emily Lambert" "Eric Bolling" "Erin Burnett"
# [25] "Francesca Vietor" "Glenn Beck" "Greg Gutfeld" "Griff Jenkins"
# [29] "Harris Kimberley Faulkner" "Harvey Weinstein" "Henry Winkler" "Hillary Hahn"
# [33] "Hoda Kotb" "Hugh Hewitt" "Jeff Ruby" "Jimmy Fallon"
# [37] "Jimmy Kimmel" "Joe Concha" "Joe Scarborough" "John Castellani"
# [41] "Jon Meacham" "Jon Stewart" "Juan Williams" "Juliet Schor"
# [45] "Ken Langone" "Kevin O’Leary" "Larry Hanael" "Larry Kudlow"
# [49] "Laura Ingraham" "Lauren Duca" "Laurie David" "Leland Vittert (?)"
# [53] "Linda Sanchez" "Liz Claman" "Mark Levin" "Michael Garcia"
# [57] "Mike Cox" "Norman Yee" "Paula Poundstone" "Petra DeJesus"
# [61] "Phil Angelides" "Piers Morgan" "Rich Hillis" "Rick Swig"
# [65] "Ricky Gervais" "Ryan Seacrest" "Sally Kohn" "Scott Turow"
# [69] "Sean Hannity" "Seth Meyers" "Stacey Abrams" "Stephen Colbert"
# [73] "Stephen Dubner" "Steve Adubato" "Stevie Nelson (?)" "Tara Setmayer"
# [77] "Tim Kaine" "Tom Cochran" "Trevor Noah" "Walter Kamau Bell"
# > levels(smallTable$role)
# [1] "activist" "actor" "anchor" "audience" "author" "caller"
# [7] "CEO" "chairman" "CNN moderator" "commisioner" "commissioner" "Commissioner"
# [13] "contestant" "environment agent" "expert" "film producer" "football player" "FOX moderator"
# [19] "guest" "historian" "host" "journalist" "moderator" "politician"
# [25] "Randy Quaids wife" "reporter" "school director" "show host" "violinist"
#
```
# Combine data into raw_df
```{r prepare_raw_df, warning=FALSE}
##### Combine all data
# raw_df <- left_join(fixTimedF0_df, f0_smooth_df, by = c("file", "t"))
# raw_df <- left_join(fixTimedF0_df, intensity_df, by = c("file", "t"))
# raw_df <- left_join(raw_df, pitchObject_df, by = c("file", "t"))
# raw_df <- left_join(raw_df, smallTable, by = "file")
raw_df <- left_join(fullTimed_df, smallTable, by = c("file"))
raw_df <- mutate(
group_by(raw_df, file, syll_label),
syll_label = ifelse(Auxiliary=="could" & syll_label=="can", "could", ifelse(Auxiliary=="may" & syll_label=="can", "may", syll_label))
)
```
# Write the raw_df table
```{r write_raw_df, warning=FALSE}
##### Write the raw data
write.csv(raw_df, "data_tables/raw_df.csv", row.names=FALSE)
```