Song Editing without Music generated by Musical Instruments.

Special Interest Groups Audio Processing
1

en.wikipedia.org/wiki/Sound‎
en.wikipedia.org/wiki/Sound_recording_and_reproduction
youtube.com/watch?v=CRzjcb22yZA
en.wikipedia.org/wiki/Digital_audio_editor
en.wikipedia.org/wiki/Plug-in_(computing)
avid.com/US/products/pro-tools-software
apps.avid.com/campaigns/2013/pro-tools-11-demo/
freegplsoftware.com/index.php/audio-video/audacity-a-free-audio-recording
audacityteam.org/
audacityteam.org/download/source
code.google.com/p/audacity/source/browse/audacity-src/trunk#trunk%2Fsrc
independentrecording.net/irn/resources/freqchart/main_display.htm
renegademinds.com/Products/GuitarDrumTrainer/FilteringInstrumentsfromSongs/tabid/111/Default.aspx
cco.caltech.edu/~boyk/spectra/spectra.htm
cs.cmu.edu/~rbd/


Is it possible to edit solo or duet song without Music generated by Musical Instruments but will have only Human Voice ?.

Example :

Bee Gees, Staying alive Song.

Youtube Video

youtube.com/watch?v=SKdVq_vNAAI

Example :

Bee Gees Brothers are singing this song. In the background, music
is being played which was generated by musical instruments viz guitar,drum beats etc. The Music played by musical instruments are in sync with the Bee Gees voice. The complete song is recorded on magnetic, Optical media viz CD, DVD, Tapes etc

Is it possible using sound editing software to edit the song without any music generated by musical instrument but the song will be only with Human beings voice i.e. Bee Gees Brothers ?.

i.e. It should be like Bee Gees Brothers are singing this song but without any musical instruments.

Example

A Song of 5 minutes duration. In the 5 minute song track there will be only :

a. Sound generated by Singer in that 5 minute song without the sound
generated by musical instrument/s.
b. Sound generated by musical instrument/s in that 5 minute song without the sound
generated by Singer.
c. Sound mixed with both by the Singer and Musical instrument/s in sync.

Requirement

The 5 minute song duration which was recorded now should only have Singer’s
Voice/Sound and not the sound generated by musical instruments. So wherever in the 5 duration only sound generated by Singer should remain, wherever in the 5 duration only sound generated by musical instrument should be eliminated and wherever in the 5 duration sound generated by both musical instrument and singer were mixed with synchronization should be that the sound generated by singer should not be eliminated but sound generated by musical instrument/s should be eliminated.

The difficult part in the sound editing software application could be editing the Sound mixing of the Singer and the Musical instruments viz guitar, piano, drum beats etc in synchronization of the song.

  1. Will Audacity software suffice the above requirements ?.

  2. A Song english,german,french,italian, hindi,marathi,tamil, gujarathi, Bengali, Punjabi etc will have three versions.

a. Music + Vocals.
b. Vocals only.
c. Music only.

The same song if heard in three different versions will have different sound effects to us. Among these the combination of both music + vocals will give us the most effective among the three.

Can Audacity software give me three versions of the same song as output disk files created with input as Vocals + Music imported audio file ?.

In that case there will be three versions 1. Vocals + Music 2. Only Music 3. Only Vocals. of the song which the consumers may like to buy such a
audio CD/DVD in the Market.

As a example :

A Audio CD/DVD having these six songs sung by Legendary Singers.

1 Michael Jackson - Thriller.
2 Phil Collins - Another Day in Paradise.
3. Stevie Wonder - I Just Called to Say I Love You.
4. Madonna - Like A Prayer.
5. Sir Elton John - Sacrifice.
6. Lionel Richie - Hello.

Now these CD/DVD will have 18 songs (6*3). The consumers may like to buy these type of Audio CD/DVD.

  1. Audacity Application changes.

Application User when executes the Audacity under windows application executable (.exe) will have two command buttons.

  1. Music Only 2. Vocals only. displayed in the Windows form. The user will import the audio file (.aiff,.mpg,.wav etc) which is a combination of
    Music + Vocals from the Local Hard drive (C:, D:, E:) using the file upload control. The imported audio file will be displayed in the
    Listbox control. The user selects the specific audio file by clicking the listbox control item. User when clicks on Music only command button, the application
    will process the request and generate Music only file on the Hard drive directory where the audio files were imported. This audio file will have only music
    generated by musical instruments eliminating the vocals. User when clicks on Vocals only command button, the application will process the request and generate Vocals only file on the Hard drive directory where the audio files were imported. This audio file will have only vocals eliminating the music generated by musical instruments.

Is it possible to build the audacity application with these features ?. I feel the Vocals only processing logic would be more complex because there are so many musical instruments viz Guitar, Drum beats, Piano, Trumpet, Saxophone, Violin etc mixed with the Human Voice in synchronization.

Thanks & Regards,
Prashant S Akerkar

2

You are leaping ahead of current technology.
“un-mixing” a track to separate out different sound sources (vocal or instrument) is difficult and often impossible. As this is the basis of your proposal I’d suggest you look at the current status of vocal removal/isolation: http://wiki.audacityteam.org/wiki/Vocal_Removal

3

Also see:
http://wiki.audacityteam.org/wiki/Proposal_Source_Separation
http://wiki.audacityteam.org/wiki/Proposal_Audio_Diff .


Gale

4

That’s currently the maximum you can expect:

I am simply switching between center, original and instruments.
It is nearly impossible to eliminate the drums from the lead track and artificial reverb will always be present in the instrument track.
I’ve also tried some other titles from above. The Bee Gees example is really a worst case. The voices are all over the place. Of course, the material was not ideal because it originates from You Tube.

5

I agree with you.

It might be difficult to remove completely the music generated by musical instruments only giving the Vocals in the audio track.
This is primarily because there are many musical instruments viz Guitar, Piano, Drum beats, Violin, Saxophone, Trumpet etc which may be
playing with the Human Voice in sync in the particular audio track. i.e. Mixing of Human voice with music generated by musical instruments
sound frequencies. One has to study every musical instrument generating different sound frequencies.

A team of musicians and singers will be required for your assistance. We know there are some legends who are both singers as well as music directors.
Example : Michael Jackson, Sir Elton John, Lionel Richie, Stevie Wonder etc

For example :

Artist A expert in Piano.
Artist B expert in Guitar.
Artist C expert in Drum Beats.
Artist D expert in Trumpet.
Artist E expert in Saxophone.
Artist F expert in Flute.

.
.
.
.

Some Music Artists may be experts knowing more than one instrument to play for example Drum Beats+Guitar, Guitar+Trumpet+Saxophone, Piano+Guitar+Drum Beats+Trumpet etc.

Example

Sir Elton John singing the song “Sacrifice”.

http://www.youtube.com/watch?v=RLFJARmnBvg

I feel the above song has only the piano being played by Sir Elton John.

  1. Start with those audio tracks where only one musical instrument is being played. For example Piano. Take the assistance of the Artist who is
    expert in playing Piano. Try removing the music generated keeping only the vocals in only those audio tracks.

    \

  2. Start with those audio tracks where only two musical instruments are being played. For example Guitar, Drum Beats. Take the assistance of the Artist who is expert in playing Guitar, Drum Beats. Try removing the music generated keeping only the vocals in only those audio tracks.

    \

  3. Start with those audio tracks where only three musical instruments are being played. For example Guitar, Drum Beats, Trumpet. Take the assistance of the Artist who is expert in playing Guitar, Drum Beats, Trumpet. Try removing the music generated keeping only the vocals in only those audio tracks.

  4. Finally Start with those audio tracks where more than one musical instruments are being played in sync with the Human Voice. For example Guitar, Drum Beats with Human Voice in Sync. Take the assistance of the Artist/s who is/are expert who knows those musical instruments. Try removing the music generated keeping only the vocals in only those audio tracks.

The 4th Type of audio tracks will be the most difficult while the 1st Type of Audio Track will be the easiest for keeping the vocals eliminating the
music generated by musical instrument/s.

i.e. Starting with easy audio tracks (Having less musical instruments) and then slowly and gradually moving to difficult audio tracks (Having more musical instruments). Identifying and segregating those audio tracks with total number of musical instruments being played in that audio track
could be the starting point for this approach.

Do you feel the above approach can help to some extend ?.

Thanks & Regards,
Prashant S Akerkar

6

I am a proficient guitar player but this won’t me let extract the guitar from a mixed track.
It’s a question of physics, maths, statistics and some braining, nothing else.