Posit AI Weblog: Audio classification with torch

Posit AI Weblog: Audio classification with torch

Variations on a theme

Easy audio classification with Keras, Audio classification with Keras: Trying nearer on the non-deep studying components, Easy audio classification with torch: No, this isn’t the primary put up on this weblog that introduces speech classification utilizing deep studying. With two of these posts (the “utilized” ones) it shares the overall setup, the kind of deep-learning structure employed, and the dataset used. With the third, it has in widespread the curiosity within the concepts and ideas concerned. Every of those posts has a distinct focus – must you learn this one?

Properly, in fact I can’t say “no” – all of the extra so as a result of, right here, you might have an abbreviated and condensed model of the chapter on this matter within the forthcoming e book from CRC Press, Deep Studying and Scientific Computing with R torch. By the use of comparability with the earlier put up that used torch, written by the creator and maintainer of torchaudio, Athos Damiani, important developments have taken place within the torch ecosystem, the top end result being that the code obtained loads simpler (particularly within the mannequin coaching half). That mentioned, let’s finish the preamble already, and plunge into the subject!

Inspecting the info

We use the speech instructions dataset (Warden (2018)) that comes with torchaudio. The dataset holds recordings of thirty completely different one- or two-syllable phrases, uttered by completely different audio system. There are about 65,000 audio information general. Our job will likely be to foretell, from the audio solely, which of thirty doable phrases was pronounced.


ds <- speechcommand_dataset(
  root = "~/.torch-datasets", 
  url = "speech_commands_v0.01",
  obtain = TRUE

We begin by inspecting the info.

[1]  "mattress"    "hen"   "cat"    "canine"    "down"   "eight"
[7]  "5"   "4"   "go"     "comfortable"  "home"  "left"
[32] " marvin" "9"   "no"     "off"    "on"     "one"
[19] "proper"  "seven" "sheila" "six"    "cease"   "three"
[25]  "tree"   "two"    "up"     "wow"    "sure"    "zero" 

Choosing a pattern at random, we see that the data we’ll want is contained in 4 properties: waveform, sample_rate, label_index, and label.

The primary, waveform, will likely be our predictor.

pattern <- ds[2000]
[1]     1 16000

Particular person tensor values are centered at zero, and vary between -1 and 1. There are 16,000 of them, reflecting the truth that the recording lasted for one second, and was registered at (or has been transformed to, by the dataset creators) a price of 16,000 samples per second. The latter data is saved in pattern$sample_rate:

[1] 16000

All recordings have been sampled on the identical price. Their size virtually at all times equals one second; the – very – few sounds which can be minimally longer we will safely truncate.

Lastly, the goal is saved, in integer kind, in pattern$label_index, the corresponding phrase being obtainable from pattern$label:

[1] "hen"
[ CPULongType{} ]

How does this audio sign “look?”


df <- data.frame(
  x = 1:length(pattern$waveform[1]),
  y = as.numeric(pattern$waveform[1])

ggplot(df, aes(x = x, y = y)) +
  geom_line(measurement = 0.3) +
      "The spoken phrase "", pattern$label, "": Sound wave"
  ) +
  xlab("time") +
  ylab("amplitude") +
The spoken word “bird,” in time-domain representation.

What we see is a sequence of amplitudes, reflecting the sound wave produced by somebody saying “hen.” Put in another way, we now have right here a time collection of “loudness values.” Even for specialists, guessing which phrase resulted in these amplitudes is an unimaginable job. That is the place area information is available in. The knowledgeable might not have the ability to make a lot of the sign on this illustration; however they could know a strategy to extra meaningfully signify it.

Two equal representations

Think about that as an alternative of as a sequence of amplitudes over time, the above wave had been represented in a approach that had no details about time in any respect. Subsequent, think about we took that illustration and tried to recuperate the unique sign. For that to be doable, the brand new illustration would in some way must include “simply as a lot” data because the wave we began from. That “simply as a lot” is obtained from the Fourier Remodel, and it consists of the magnitudes and part shifts of the completely different frequencies that make up the sign.

How, then, does the Fourier-transformed model of the “hen” sound wave look? We receive it by calling torch_fft_fft() (the place fft stands for Quick Fourier Remodel):

dft <- torch_fft_fft(pattern$waveform)
[1]     1 16000

The size of this tensor is identical; nevertheless, its values are usually not in chronological order. As an alternative, they signify the Fourier coefficients, akin to the frequencies contained within the sign. The upper their magnitude, the extra they contribute to the sign:

magazine <- torch_abs(dft[1, ])

df <- data.frame(
  x = 1:(length(pattern$waveform[1]) / 2),
  y = as.numeric(magazine[1:8000])

ggplot(df, aes(x = x, y = y)) +
  geom_line(measurement = 0.3) +
      "The spoken phrase "",
      "": Discrete Fourier Remodel"
  ) +
  xlab("frequency") +
  ylab("magnitude") +
The spoken word “bird,” in frequency-domain representation.

From this alternate illustration, we might return to the unique sound wave by taking the frequencies current within the sign, weighting them in accordance with their coefficients, and including them up. However in sound classification, timing data should absolutely matter; we don’t actually wish to throw it away.

Combining representations: The spectrogram

The truth is, what actually would assist us is a synthesis of each representations; some kind of “have your cake and eat it, too.” What if we might divide the sign into small chunks, and run the Fourier Remodel on every of them? As you might have guessed from this lead-up, this certainly is one thing we will do; and the illustration it creates known as the spectrogram.

With a spectrogram, we nonetheless hold some time-domain data – some, since there’s an unavoidable loss in granularity. Then again, for every of the time segments, we study their spectral composition. There’s an essential level to be made, although. The resolutions we get in time versus in frequency, respectively, are inversely associated. If we cut up up the alerts into many chunks (known as “home windows”), the frequency illustration per window is not going to be very fine-grained. Conversely, if we wish to get higher decision within the frequency area, we now have to decide on longer home windows, thus shedding details about how spectral composition varies over time. What appears like an enormous downside – and in lots of instances, will likely be – gained’t be one for us, although, as you’ll see very quickly.

First, although, let’s create and examine such a spectrogram for our instance sign. Within the following code snippet, the dimensions of the – overlapping – home windows is chosen in order to permit for cheap granularity in each the time and the frequency area. We’re left with sixty-three home windows, and, for every window, receive 2 hundred fifty-seven coefficients:

fft_size <- 512
window_size <- 512
energy <- 0.5

spectrogram <- transform_spectrogram(
  n_fft = fft_size,
  win_length = window_size,
  normalized = TRUE,
  energy = energy

spec <- spectrogram(pattern$waveform)$squeeze()
[1]   257 63

We will show the spectrogram visually:

bins <- 1:dim(spec)[1]
freqs <- bins / (fft_size / 2 + 1) * pattern$sample_rate 
log_freqs <- log10(freqs)

frames <- 1:(dim(spec)[2])
seconds <- (frames / dim(spec)[2]) *
  (dim(pattern$waveform$squeeze())[1] / pattern$sample_rate)

image(x = as.numeric(seconds),
      y = log_freqs,
      z = t(as.matrix(spec)),
      ylab = 'log frequency [Hz]',
      xlab = 'time [s]',
      col = hcl.colors(12, palette = "viridis")
important <- paste0("Spectrogram, window measurement = ", window_size)
sub <- "Magnitude (sq. root)"
mtext(aspect = 3, line = 2, at = 0, adj = 0, cex = 1.3, important)
mtext(aspect = 3, line = 1, at = 0, adj = 0, cex = 1, sub)
The spoken word “bird”: Spectrogram.

We all know that we’ve misplaced some decision in each time and frequency. By displaying the sq. root of the coefficients’ magnitudes, although – and thus, enhancing sensitivity – we had been nonetheless in a position to receive an affordable end result. (With the viridis shade scheme, long-wave shades point out higher-valued coefficients; short-wave ones, the other.)

Lastly, let’s get again to the essential query. If this illustration, by necessity, is a compromise – why, then, would we wish to make use of it? That is the place we take the deep-learning perspective. The spectrogram is a two-dimensional illustration: a picture. With pictures, we now have entry to a wealthy reservoir of methods and architectures: Amongst all areas deep studying has been profitable in, picture recognition nonetheless stands out. Quickly, you’ll see that for this job, fancy architectures are usually not even wanted; an easy convnet will do an excellent job.

Coaching a neural community on spectrograms

We begin by making a torch::dataset() that, ranging from the unique speechcommand_dataset(), computes a spectrogram for each pattern.

spectrogram_dataset <- dataset(
  inherit = speechcommand_dataset,
  initialize = perform(...,
                        pad_to = 16000,
                        sampling_rate = 16000,
                        n_fft = 512,
                        window_size_seconds = 0.03,
                        window_stride_seconds = 0.01,
                        energy = 2) {
    self$pad_to <- pad_to
    self$window_size_samples <- sampling_rate *
    self$window_stride_samples <- sampling_rate *
    self$energy <- energy
    self$spectrogram <- transform_spectrogram(
        n_fft = n_fft,
        win_length = self$window_size_samples,
        hop_length = self$window_stride_samples,
        normalized = TRUE,
        energy = self$energy
  .getitem = perform(i) {
    merchandise <- tremendous$.getitem(i)

    x <- merchandise$waveform
    # be sure all samples have the identical size (57)
    # shorter ones will likely be padded,
    # longer ones will likely be truncated
    x <- nnf_pad(x, pad = c(0, self$pad_to - dim(x)[2]))
    x <- x %>% self$spectrogram()

    if (is.null(self$energy)) {
      # on this case, there's a further dimension, in place 4,
      # that we wish to seem in entrance
      # (as a second channel)
      x <- x$squeeze()$permute(c(3, 1, 2))

    y <- merchandise$label_index
    list(x = x, y = y)

Within the parameter listing to spectrogram_dataset(), notice energy, with a default worth of two. That is the worth that, until instructed in any other case, torch’s transform_spectrogram() will assume that energy ought to have. Beneath these circumstances, the values that make up the spectrogram are the squared magnitudes of the Fourier coefficients. Utilizing energy, you possibly can change the default, and specify, for instance, that’d you’d like absolute values (energy = 1), another constructive worth (similar to 0.5, the one we used above to show a concrete instance) – or each the actual and imaginary components of the coefficients (energy = NULL).

Show-wise, in fact, the total advanced illustration is inconvenient; the spectrogram plot would want a further dimension. However we might effectively ponder whether a neural community might revenue from the extra data contained within the “complete” advanced quantity. In any case, when lowering to magnitudes we lose the part shifts for the person coefficients, which could include usable data. The truth is, my checks confirmed that it did; use of the advanced values resulted in enhanced classification accuracy.

Let’s see what we get from spectrogram_dataset():

ds <- spectrogram_dataset(
  root = "~/.torch-datasets",
  url = "speech_commands_v0.01",
  obtain = TRUE,
  energy = NULL

[1]   2 257 101

We have now 257 coefficients for 101 home windows; and every coefficient is represented by each its actual and imaginary components.

Subsequent, we cut up up the info, and instantiate the dataset() and dataloader() objects.

train_ids <- sample(
  measurement = 0.6 * length(ds)
valid_ids <- sample(
  measurement = 0.2 * length(ds)
test_ids <- setdiff(
  union(train_ids, valid_ids)

batch_size <- 128

train_ds <- dataset_subset(ds, indices = train_ids)
train_dl <- dataloader(
  batch_size = batch_size, shuffle = TRUE

valid_ds <- dataset_subset(ds, indices = valid_ids)
valid_dl <- dataloader(
  batch_size = batch_size

test_ds <- dataset_subset(ds, indices = test_ids)
test_dl <- dataloader(test_ds, batch_size = 64)

b <- train_dl %>%
  dataloader_make_iter() %>%

[1] 128   2 257 101

The mannequin is a simple convnet, with dropout and batch normalization. The true and imaginary components of the Fourier coefficients are handed to the mannequin’s preliminary nn_conv2d() as two separate channels.

mannequin <- nn_module(
  initialize = perform() {
    self$options <- nn_sequential(
      nn_conv2d(2, 32, kernel_size = 3),
      nn_max_pool2d(kernel_size = 2),
      nn_dropout2d(p = 0.2),
      nn_conv2d(32, 64, kernel_size = 3),
      nn_max_pool2d(kernel_size = 2),
      nn_dropout2d(p = 0.2),
      nn_conv2d(64, 128, kernel_size = 3),
      nn_max_pool2d(kernel_size = 2),
      nn_dropout2d(p = 0.2),
      nn_conv2d(128, 256, kernel_size = 3),
      nn_max_pool2d(kernel_size = 2),
      nn_dropout2d(p = 0.2),
      nn_conv2d(256, 512, kernel_size = 3),
      nn_adaptive_avg_pool2d(c(1, 1)),
      nn_dropout2d(p = 0.2)

    self$classifier <- nn_sequential(
      nn_linear(512, 512),
      nn_dropout(p = 0.5),
      nn_linear(512, 30)
  ahead = perform(x) {
    x <- self$options(x)$squeeze()
    x <- self$classifier(x)

We subsequent decide an appropriate studying price:

mannequin <- mannequin %>%
    loss = nn_cross_entropy_loss(),
    optimizer = optim_adam,
    metrics = list(luz_metric_accuracy())

rates_and_losses <- mannequin %>%
rates_and_losses %>% plot()
Learning rate finder, run on the complex-spectrogram model.

Based mostly on the plot, I made a decision to make use of 0.01 as a maximal studying price. Coaching went on for forty epochs.

fitted <- mannequin %>%
    epochs = 50, valid_data = valid_dl,
    callbacks = list(
      luz_callback_early_stopping(endurance = 3),
        max_lr = 1e-2,
        epochs = 50,
        steps_per_epoch = length(train_dl),
        call_on = "on_batch_end"
      luz_callback_model_checkpoint(path = "models_complex/"),
    verbose = TRUE

Fitting the complex-spectrogram model.

Let’s examine precise accuracies.


With thirty lessons to tell apart between, a last validation-set accuracy of ~0.94 appears to be like like a really respectable end result!

We will verify this on the check set:

consider(fitted, test_dl)
loss: 0.2373
acc: 0.9324

An attention-grabbing query is which phrases get confused most frequently. (After all, much more attention-grabbing is how error possibilities are associated to options of the spectrograms – however this, we now have to depart to the true area specialists. A pleasant approach of displaying the confusion matrix is to create an alluvial plot. We see the predictions, on the left, “movement into” the goal slots. (Goal-prediction pairs much less frequent than a thousandth of check set cardinality are hidden.)

Alluvial plot for the complex-spectrogram setup.


That’s it for immediately! Within the upcoming weeks, anticipate extra posts drawing on content material from the soon-to-appear CRC e book, Deep Studying and Scientific Computing with R torch. Thanks for studying!

Picture by alex lauzon on Unsplash

Warden, Pete. 2018. “Speech Instructions: A Dataset for Restricted-Vocabulary Speech Recognition.” CoRR abs/1804.03209. http://arxiv.org/abs/1804.03209.

Leave a Reply

Your email address will not be published. Required fields are marked *