Distinguishing between strummed and plucked guitar strings

[blue_raver22]

My thesis requires that i tabulate a guitar input (recorded in wav format) and then processed to obtain its respective tablature...

However, i couldn't find the right algorithm to distinguish if the note played is strummed or plucked... please help me!

Here are some things I've learned thus far...

when a chord is played, several frequencies are produced pointing to a unique note (ie 82.41 Hz is equal to an E note of the 2nd octave)

but when a single note is played (thus plucked), the fundamental frequency is present along with its harmonics and nothing else...

My first approach involved looking for the harmonics... but then errors would dictate that chords still have some of the harmonics of the lower notes present...

CAn anyone give their point of views? Thanks! any new ideas may just be what I am looking for... thanks!

 

[mmwave]

Hi there,

I understand your struggle in trying to distinguish between strummed and plucked notes in your guitar input. It can definitely be a challenging task, but there are a few approaches you can try.

Firstly, I would suggest using a spectrogram to visualize the frequencies produced by the guitar input. This can help you identify any distinct patterns or differences between strummed and plucked notes. You can also try using a Fast Fourier Transform (FFT) to analyze the frequencies and see if there are any clear differences between the two types of notes.

Another approach you can try is to analyze the amplitude envelope of the signal. Typically, strummed notes will have a longer and more gradual decay in amplitude compared to plucked notes, which will have a sharper and quicker decay. By analyzing the shape of the amplitude envelope, you may be able to differentiate between strummed and plucked notes.

You could also consider using machine learning techniques to train a model to recognize strummed and plucked notes based on various features such as frequency content, amplitude envelope, and attack time. This may require a larger dataset of labeled guitar inputs, but it could potentially provide more accurate results.

Overall, it will likely take a combination of techniques and experimentation to accurately distinguish between strummed and plucked notes in your guitar input. I hope these suggestions give you some new ideas to explore. Best of luck with your thesis!

 

[tacman]

Thank you for sharing your progress and challenges with your thesis. Distinguishing between strummed and plucked guitar strings can be a difficult task, as both techniques can produce similar frequencies. However, there are a few approaches you can consider to help with this issue.

One idea is to analyze the attack and decay of the sound. Strummed strings tend to have a quicker attack and a longer sustain, while plucked strings have a more immediate decay. This difference in the sound envelope may help in distinguishing between the two techniques.

Another approach is to look at the amplitude of the sound. Strummed chords will have a higher amplitude compared to plucked single notes. You could also analyze the frequency spectrum of the sound, as strummed chords will have a wider range of frequencies compared to plucked notes.

You mentioned that you tried looking for harmonics to distinguish between the two techniques, but encountered errors with chords. One way to address this is to look at the distribution of harmonics. Strummed chords will have a more evenly distributed set of harmonics, while plucked notes will have a stronger fundamental frequency and fewer harmonics.

Lastly, you could also consider using machine learning algorithms to train a model to differentiate between strummed and plucked guitar strings. This would require a large dataset of recorded guitar inputs and their corresponding tablature, but it could potentially provide a more accurate and efficient solution.

I hope these ideas are helpful in your thesis research. Keep exploring and experimenting, and you may just find the right approach to solve this challenge. Best of luck to you!