1) How are things working? Did you hit any bumps in the road? Is your plan for your project essentially the same, or are you going to go in a new direction? Talk briefly about your plan for the last two weeks of class until the project is due. (11/25/14-12/9/2014)

 

 

Our project is following the plan. After being able to successfully synthesize the sound of a user selected instrument with a user input frequency, we are working on morphing multiple notes together with a more smooth transition in between. Audio morphing is not as easy to implement as we originally thought however and will require a concentrated effort to complete. For the coming two weeks, we are planning to develop our algorithm for morphing different notes together to yield a small chunk of melody. We also want to integrate the morphed audio with the GUI so that for each selection of instrument, a short clip of well morphed audio of selected instrument with the customized frequency would be reconstructed and played.

 

 

2) What in-class DSP tools are you incorporating into your project?

 

 

We are incorporating the Discrete Fourier Transform to analyze musical notes, and implementing the DFT through the Fast Fourier Transform in Matlab. We are also using the Short Time Fourier Transform to assist in our spectral morphing of one note from another. These two things fall in the realm of frequency domain analysis. We are also using music processing techniques like the ones briefly covered in one of the weekly homework assignments.

 

 

3) What out-of-class DSP tools are you incorporating?

 

 

We are using audio morphing to provide a more smooth transition from one note to another. There are several ways of morphing one note into another. One of the simplest way to morph would be the idea of cross-fading of amplitude in the time domain. Basically, it is just to fade out the amplitude of the first note and gradually fade into the amplitude of the second note. We can add or overlap the two fading phases together to make the morphing.

 

Another way to morph the notes would be the manipulation in the frequency domain. We are still exploring different ways in achieving morphing. But one way would be to cross-synthesize two notes. Cross-synthesize is the idea that we impress the spectral envelope of one note on the flattened spectrum of another. In order to achieve cross synthesis, we have to perform short-time Fourier Transform, extract the spectral envelope and flatten the carrier’s spectrum. Then we would multiply the flattened spectral envelope of carrier with the corresponding modulator frame. Then we would be able to successfully make the transition between two notes.

 

4) What is the coolest / most fun thing you've done on your project so far?

 

The most interesting thing we’ve done so far has got to be the use of musical analysis to synthesize a musical note from the harmonic frequencies of the note. It is nice to see our fourier analysis techniques and frequency domain analysis translate into a recognizable tone.

Another fun aspect of the project so far has been hearing the varied sounds we’ve produced, whether accidentally or intentionally. It’s so convenient to have sound has a medium for understanding the complex algorithms and tools from class, and this often has humorous effects.  For instance, while working on the audio morphing component we have produced a variety of sounds, some pleasant and some not so much.  Hearing and making sense of what composes a pleasing sound as compared to a harsh sound is both fascinating and stimulating!