Audio Recording

There might be occasions where you really need to produce a karaoke version of a certain song. The primary reason is that someone will like to sing along with it. There are lots of quality audio recording software that can be used to accomplish this task. For example, if you are using Adobe Audition you can use read this tutorial on creating karaoke or removing vocals in Adobe Audition.

You can as well use open source software such as Ardour or Audacity. However, once you have succeeded in isolating the vocal/speech tracks from the track; the results are not actually perfect. There are some minor audio quality issues on it. For example, you might notice that there are still some vocals on it. Or the sound quality is very poor. Other cases are that the sections where you have removed the speech/voices are substantially low in amplitude and could not be properly reproduced in other audio players. Examples are cellular phones where the frequency range is limited. This is the situation of one of the site readers asking for some advice:

Hi there,
I am a beginner with audio recording and mixing, and I have an at2020. But my problem is that I’m gonna record a song with my friend, and I have the official karaoke, it’s from a Barbie movie where you have an original karaoke on the dvd, but the karaoke, is not fully instrumental, only the singing is instrumental, the spoken parts are still there. So I tried to remove the vocals with a basic karaoke effect in Ardour and succeeded, it sounded super nice and with fade in and fade out and some gain, I made it sound like it was originally like that. But when I put it on my phone, just to have it easy to rehearse, the parts I removed the vocals where super low, but if I put headphones on my phone I would hear it good again, so the problem is just when I put it on speakers… And on my computers speakers there is not any problem or difference… Do you know how I could fix this? Thanks for at least reading my email, and I hope you know a solution to my problem, because of course, it won’t be a problem if it sounds good with headphone, but I’d like it to be as best as it can be… The effect is applied around 20 seconds… That is where the problem is, and some other places have the effect to…

First, let’s compare how it sounds when played in a cellular phone and with headphones. Cellular phone speakers have limited frequency range around 300Hz to 4000Hz. Using Adobe Audition FFT filter, the audio can be simulated as follows:

This is how it sounds like:

The problem is around 20 seconds; you will strongly hear a substantial drop in volume/amplitude at 20 seconds of the audio sample. Now let’s listen to an unaltered audio when played at full spectrum of the frequency range.

The audio drop is not as noticeable as before. At 20 seconds, the portion where you have removed the speech parts did not have sufficient gain at 300Hz to 4000Hz. Examining further the problem, it seems like an “unusual” 4dB drop in the audio around 20 seconds such as shown in the screenshot below:

This is the specific section of the audio without the gain applied to the “low volume” sections.

Select those sections with low volume and apply a 4dB gain (simple amplification of audio by 4dB).

This is how it sounds like after 4dB amplification on affected region:

As long as it sounds very good during actual performance (which is played using a full spectrum studio monitors), I won’t worry about it too much.