Questions tagged [audio]
The audio tag has no usage guidance.
18 questions
0
votes
0
answers
54
views
Good training/test results, but very poor performance in inference as the stream data is coming
I am interested in the audio classification problem. After labeling the audio recordings I have in Praat software environment, I extract the MFCC features from each labeled frame and create an SVM ...
- 23
0
votes
1
answer
39
views
How to identify the middle of the first near-zero trough in a graph?
I have audio files and timestamped fragment transcriptions that I have force-aligned with third party software.
Unfortunately the software places the end of each fragment right before the start of the ...
- 101
2
votes
1
answer
49
views
Minuscule data for classification problem
I am working on a frog classification problem where I have 15 unique species and 2 recordings per species. Would it be a valid experiment if I were to have my training set of the 15 unique species ...
- 21
0
votes
1
answer
237
views
RNN/LSTM networks on spectrograms underfitting massively - is the CNN encoder a prerequisite?
I am prototyping a pipeline on the FSDD dataset (audio/10-class classification); the audio data are loaded with librosa, 0-padded/trimmed to 0.5 sec (4000-dimensioned numpy vectors) each and converted ...
- 81
1
vote
1
answer
66
views
"No Classification" for a CNN Neural Network on Audio
It it possible for a Convolutional Neural Network to output "no classification" for a multi-audio classifier? Is it common practice to set % thresholding to the output vector of ...
0
votes
1
answer
57
views
K-fold Stratified cross validation on a dataset with examples of variable length
I have a dataset of audio recording of variable length with large std which are heavily imbalanced in terms of total duration per class. So in stratified k-fold cross validation I would like to main ...
- 11
0
votes
1
answer
399
views
What is Hamming Window in Audio Analysis?
I am reading this paper. The paper writes
... an input waveform of t seconds is converted into a sequence of 128-D log mel filter bank features computed with 25 ms Hamming window every 10 s. The ...
- 21
0
votes
1
answer
651
views
How does the plot of the output of librosa.stft work?
So, I have been analyzing audio data. I used the librosa package and used its .stft() function to calculate the Fourier transform of my audio data. I did the following-
...
2
votes
1
answer
2k
views
Looking for repeated patterns in time series data
I have spent the best part of the last few days searching forums and reading papers trying to solve the following question. I have thousands of time series arrays each of varying lengths containing a ...
1
vote
0
answers
2k
views
Why almost all neural speech processing involves Mel Spectrograms?
What are the reasons behind almost all speech processing whether it be generative or recognition heavily based on Mel Spectrograms?
In a conversation with a signal processing expert I was asked why ...
- 111
0
votes
0
answers
38
views
Is it a good practice to pad signal before feature extraction?
I have a question for you - is padding, before feature extraction with VGGish, a good practice? Our padding technique is to find the longest signal (which is loaded .wav signal) and then in every ...
- 1
2
votes
0
answers
85
views
goodness of fit for psychometric data (perceptual threshold)
I'm running an experiment on perceptual thresholds in audio. I'll try not to bog you down with too many details:
The experiment is about vibrato speed; specifically, when can you tell the difference ...
- 21
1
vote
0
answers
39
views
What is the name of this data denoising method
I've been working on extracting data from an extremely noisy signal. The signal itself is the 1st derivative of raw mean squared (RMS) of an audio that may contain segments with some single low ...
- 11
4
votes
1
answer
852
views
WGAN-GP stability loss
I am training a Conditional WaveGAN (1D DCGAN for audio) using WGAN-GP whose generator is of an auotencoder architecture. The network is trained to take an audio input, compress it, then decompress it ...
- 249
2
votes
1
answer
2k
views
What are good basic loss functions for audio generation? (TTS)
I'm planning to make an audio generation NN.
While I'm reasonably ok with neural networks in general, wavenets, etc., something is not quite clear.
What are good loss functions for audio, considering ...
- 121
1
vote
1
answer
2k
views
Conv2D Kernel size for audio-related tasks
So I've been working on this audio-rec task for a while now, and I've had some good luck using 2D convolutions on the spectrogram of audio (I've also tried Mel-spectrograms, the difference is minor in ...
2
votes
1
answer
50
views
Different fonts of audio from a single audio source
I know that is a duplicated question. But there is no answer for the original one and I want something more specific. On original question, the user Caaarlos wants the interpret different fonts of ...
- 53
0
votes
0
answers
146
views
Benchmarking model in speech recognition with different language
My supervisor asked me to benchmark my method in classifying speech signal with other language. I am doing Malay language speech recognition. To benchmark my method/feature used, I need to test ...
- 21