To my knowledge, a gaussian window is the optimal tradeoff between time and frequency resolution. Other window functions favour one over the other. See Wikipedia
@@Eigensteve if you just want to know when frequencies occur at what time..dp you need any transform at all? Wouldn't you just look at the original graph of frequency versus time? you wouldnt need a gabor or any other kind of transform for that matter
Note the spectrogram function is part of the Signal Processing Toolbar add on for Matlab. So, unless you buy it, this example won't run. Ran the python version instead, but at least on my Mac, librosa won't open the mp3 without additional pip config. Not complaining, just making notes for those who hit these issues too. Love the course, btw!
I don’t want to claim to have ESP, but I just knew there was going to be a sound effect once he started to describe the chirp command. Was not disappointed. 😂 These are used a ton in radio - waterfall charts. Could be fun to check out, if anyone has ever heard of or played with cheap SDR devices. Explore some of the EM spectrum.
Hi Steve. Just to verify, the codes in lines 8 and 9 are mathematically equivalent, correct? In line 9, are you explicitly defining x as an up-chip interpolated quadratically? Otherwise line 9 simply overwrites line 8. Thank you.
Thank you, Sir, for this very informative tutorial. I was studying the shazam fingerprint and, it leads me to your tutorial. Now, I fully understand the math behind it
Sorry to be off topic but does any of you know a way to log back into an instagram account? I was stupid lost the login password. I love any help you can offer me.
thank you dr for the amazing description ,I Have small question , if I want to find the Gabor transform of a sinusoidal signal which contains some distortion , I should use spectrogram command or there is a Matlab command can find Gabor transform ?
This is a good video I am trying to adapt this to how I can detect detonation in my motor by using the knock sensors audio. I have a strong powerful band similar to that that goes up with rpm and load which is the harmonics of the motor operating. Now im trying to see that a sharp knock event at 5.7hz should show up in the spectrogram as a different colored spike correct?
I guess the way to translate that would be imagine pressing a high piano key and holding it and then tapping a lower octave key it should register that spike correct?
Can we use the Gábor transform to identify a recorded voice on phones. Say the call was received from overseas. Given the equipment used,the distance, etc can we still reconnise the real voice if recorded separately. In other words can we proove the accused of making a criminal call is not guilty.
Sir how the vocal folds vibrate at multiple frequencies at the same time Actually vocal folds vibrate at specific frequency right thhen where do all harmonics come from
The voice has three parameters, amplitude, frequency and spectrum -- lungs, vocal folds, vocal tract [mouth, nose]. The vocal folds produce a form of pulse wave. The mouth / vocal tract creates resonances / resonant frequency regions, called formant frequencies. Do a Wiki search for 'vowels', and vowel formants. The vocal folds [almost always] vibrate at "one" frequency at a time. [There are some exceptions, eg Janice Joplin and other singers who produce vocal 'sizzle'. This is not the same technique as Tuwan and Tibetan 'harmonic singing'.
Excuse my lack of knowledge and terminology. In this example there is clearly a 'trace' we see a smooth transition from a lower frequency to a higher frequency. Suppose we wanted to extract only this 'trace' how would one go about this?
There are different ways to produce visualizations of spectrum over time. The explanation shown here is the inner workings of how the signal is converted from the time domain to the frequency domain. Spectrograms and mp3 files use the same basic technique for this transformation.
![Spectrogram Examples [Python]](http://i.ytimg.com/vi/TJGlxdW7Fb4/mqdefault.jpg)
![Spectrogram Examples [Python]](/img/tr.png)
This book and lecture series are proving to be very helpful 👍
What role does the Gaussian play? When should one consider other window functions?
Good quesiton, I would also very much like to know more about this! :)
To my knowledge, a gaussian window is the optimal tradeoff between time and frequency resolution. Other window functions favour one over the other.
See Wikipedia
Yes, you can use other windows, but @jamen1993 has the right idea.
@@Eigensteve if you just want to know when frequencies occur at what time..dp you need any transform at all? Wouldn't you just look at the original graph of frequency versus time? you wouldnt need a gabor or any other kind of transform for that matter
I would find a video about your take on the constant Q transform quite interesting. It would certainly be an excellent follow-up to this one.
Note the spectrogram function is part of the Signal Processing Toolbar add on for Matlab. So, unless you buy it, this example won't run. Ran the python version instead, but at least on my Mac, librosa won't open the mp3 without additional pip config. Not complaining, just making notes for those who hit these issues too. Love the course, btw!
Why do you need the gabor transform at all? Woudlnt you just look at the original graph of time versus freqeuncy?
Nicely explained and beautifully illustrated. Thank you for pointing out the book! ... plus access to pdf. Will dive into it.
Your book is awsome!! I love that simple way with which your explaining things.
Thanks so much!
I don’t want to claim to have ESP, but I just knew there was going to be a sound effect once he started to describe the chirp command. Was not disappointed. 😂
These are used a ton in radio - waterfall charts. Could be fun to check out, if anyone has ever heard of or played with cheap SDR devices. Explore some of the EM spectrum.
Hi Steve. Just to verify, the codes in lines 8 and 9 are mathematically equivalent, correct? In line 9, are you explicitly defining x as an up-chip interpolated quadratically? Otherwise line 9 simply overwrites line 8. Thank you.
Thank you, Sir, for this very informative tutorial. I was studying the shazam fingerprint and, it leads me to your tutorial. Now, I fully understand the math behind it
Sorry to be off topic but does any of you know a way to log back into an instagram account?
I was stupid lost the login password. I love any help you can offer me.
Very interesting, I was wondering how to get timeseries data from the FT
Can we use it for digital signals for example pressure or acceleration histories of shock waves?
thank you dr for the amazing description ,I Have small question , if I want to find the Gabor transform of a sinusoidal signal which contains some distortion , I should use spectrogram command or there is a Matlab command can find Gabor transform ?
Very Useful to us.Thanks
th-cam.com/video/3HzgUx9jdy8/w-d-xo.html
chapgpt 인터프리터로 스펙트로그램 만들기
입니다.
How can i calculate various parameters in magnetic resonance spectroscopy using matlab
Would you also use this method for real time system identification of some PDE?
This is a good video I am trying to adapt this to how I can detect detonation in my motor by using the knock sensors audio. I have a strong powerful band similar to that that goes up with rpm and load which is the harmonics of the motor operating. Now im trying to see that a sharp knock event at 5.7hz should show up in the spectrogram as a different colored spike correct?
I guess the way to translate that would be imagine pressing a high piano key and holding it and then tapping a lower octave key it should register that spike correct?
Your videos are amazing! Thanks for the explanations!
Can we use the Gábor transform to identify a recorded voice on phones. Say the call was received from overseas. Given the equipment used,the distance, etc can we still reconnise the real voice if recorded separately. In other words can we proove the accused of making a criminal call is not guilty.
you shouldn't use jet colormap exactly because of this 3:35
Good thing that this is a reversible problem, right? Otherwise the Beethoven sonata would be "de-composed" forever (8:30 min into the lecture)...
Hmmm . the Beethoven piece is not the 'sound'. The sound is a representation of the creation.
Sir how the vocal folds vibrate at multiple frequencies at the same time
Actually vocal folds vibrate at specific frequency right thhen where do all harmonics come from
The voice has three parameters, amplitude, frequency and spectrum -- lungs, vocal folds, vocal tract [mouth, nose]. The vocal folds produce a form of pulse wave. The mouth / vocal tract creates resonances / resonant frequency regions, called formant frequencies. Do a Wiki search for 'vowels', and vowel formants. The vocal folds [almost always] vibrate at "one" frequency at a time. [There are some exceptions, eg Janice Joplin and other singers who produce vocal 'sizzle'. This is not the same technique as Tuwan and Tibetan 'harmonic singing'.
Cool Matlab programming!
Excuse my lack of knowledge and terminology. In this example there is clearly a 'trace' we see a smooth transition from a lower frequency to a higher frequency. Suppose we wanted to extract only this 'trace' how would one go about this?
This is so cool
*SpecTROgram
this is nuts
there's a better way to get a spectrogram of a sound
There are different ways to produce visualizations of spectrum over time. The explanation shown here is the inner workings of how the signal is converted from the time domain to the frequency domain. Spectrograms and mp3 files use the same basic technique for this transformation.
thanks for the video 🤍