bare with me friends...its a dousey
Title
Physical Sound: The balance of form and content within recorded music.
Abstract
Music is a universal phenomenon, permeating most every culture throughout the world. Since the development of sound
recoding, recorded music has become integral to society and popular culture. However, many take for granted the
process that is actually occurring when they listen to music as well as the cognitive understanding of what they are actually
hearing. A song is an abstract idea, as is all music on a practical level, though in order for one to hear a song it must
travel as a pressure wave from the source, through the air, and into one's ears. Thus, the process of recording forces a
piece of music into the physical form the song takes on its travel into the listener's brain, and allows it to have the same
properties for every listen. When someone listens to a song, they are hearing the abstract ideas created by the artist via
the physical interpretation created by the recording team. The latter half of that equation is often overlooked by the average listener just as they do not realize the ultimate impact it has on their listening experience and reception of the song.
Nearly all pop songs, for example, have very similar production qualities, so much so that through analysis, one finds
certain properties more prevalent than others within the most popular of songs. In fact, many parameters (both in the
song writing realm and the production realm) can be measured and traced across the pop music charts, leading researchers to
the conclusion that certain types of sounds, feels, and textures (all purely on a sonic level) have a conclusive and definite
impact on the song's overall popularity and public reception. Therefore, once can actually measure the 'hit potential' of a
recorded song by comparing it to figures in databases generated through analyzing these parameters from song to song.
Description
The purpose of my project is to generate public awareness of the important relationship between a song's musical
content and its physical form. This end will be achieved through user interaction with my online software (which will draw conclusions based on a database that I create). The user must input his or her own choice of song as well as a goal for that song (be it popularity, innovativeness, or
popularity within one of the analyzed genres), then the program runs the song and uses the input goal to compare the song to the right data to
generate results in two forms: a physical three dimensional model of the song and a results table. The model will help
the reader connect with the idea that a recorded song indeed does have a physical form, which was (supposedly) crafted carefully for that specific song. The results table will show how that song compared to the database and how likely it will
accomplish the goal the reader set (as well as offering results to the other possible goals). The results are merely statistics based on the database and are open for interpretation. If the results are not desirable or just unexpected then there are several reasons that could be gleaned from the information. An obvious reason for displeasing results is that the user just didn’t know much about other genres of music or physical properties of music and therefore would be enlightened by knowing that the input song has properties more prevalent in other types of music they do not know about. Another reason could be that the balance between form and content is out of whack. Either the production value is high but the song content is not well suited for that type of production or vice versa. The program merely generates statistics based on purely sonic parameters and public opinion (in the form of billboard charts) and thus, has no bias. So, it does not offer advice or potential ways to fix a song, it merely demonstrates the relationship and balance between form and content of recorded music, with hopes of bringing the user to a better understanding of what they are actually listening to.
Methodology
The software program will analyze a digital music file (extracting the information that governs the waveform) based on a number of set parameters:
Length
Tempo
Key
Frequency response
Amplitude
Time signature
Time signature changes
Key Changes
Timbre
Once the program is up and running, I will go genre by genre, running 150 to 200 of the most popular songs in that genre (taking into account periods of recoding technology development) until I have a data set from each genre in question. All the data will be compared using a weighted average algorithm, finding the common points and trends throughout the music. The result will be the model song for that genre, meaning it (stereotypically and theoretically) is the ideal song—perfectly form/content balanced—in that genre. Once I have all the models then I can generate the perfect innovative song model by using the same algorithmic process on all the models to create the ideal song that sounds nothing like any of the models—hence perfectly innovative.
The genres in question:
Pop
Alternative Rock
Indie Rock
Main Stream Hip Hop
RnB
Jazz
Smooth jazz
Progressive Rock
Singer Song Writer
Folk
Metal
Timeline
Summer:
-Research existing companies and data.
-Assemble engineering team
-Research waveform parameters
-Research genres and recording technology periods.
-Assemble the lists of songs to be used.
Fall
-Begin software development.
-Assemble library of music
-Design interface.
-Finish software and begin to generate database
-Apply for grants
-Finish database
Spring
-Purchase domain name
-Create website
-Review data gathered from website
-Conclude if this project actually worked.
-Present findings
-Turn in project
Budget
There will be a number of things that will require financial assistance for this project namely acquiring all the music. Depending on my financial status I will have to use what ever format of music I can afford. I will also need to pay (or compensate some how) the team that will develop this software. I will have to purchase and maintain a domain name and website.
Subscribe to:
Post Comments (Atom)
2 comments:
There are many things I like about this project, one of this is the side-effect that by categorizing the popular songs, you can discover what the prototypical song for each genre is, as well as clearly define what aspects create a certain genre.
However, as we discussed in class, you'll have to work an algorithm that properly encompasses the songs as well as weights them according to actual popularity.
In regards to the issues concerning outliers, I would say not to focus on them too much, incorporate the successful ones into the mix, however, since real life can't predict the outliers, neither should your program. In which case, it may be best to not even incorporate the popular outliers into your program, as they would just obscure the data from the more traditional versions.
Also, try to take a look at the pre-existing programs that are similiar to this, if you can. I know you're aware of some (I believe there's something like this in Germany).
Looks like a really cool project.
Make sure you mention in your proposal your criteria for picking the database of songs. There should be some tangible reason for the choices. Perhaps pick the 100 best-selling songs in each genre or something like that.
Also, in your data, what are you going to do about keys since you can't average keys together? All of your other characteristics are numerical except for key. It is an important characteristic, but I wonder how you will average it like the others. Math would make Db the average between C and D.
Post a Comment