POD 5 Family KTV Recording System... A Customer's Review

[love_stinks]

---

Hi guys,

Thanks for the time spent on reading my post, I have personalized message to the following though:

vipvip68 - thanks for the expressed concern :mrgreen:
bfg9000 - your explanations are well noted
tolstoi - i didn't think Pod 5 is unprofessional
Edgeowns - Good for you, you had a good experience there, not everybody can be as lucky as you though
Mango - Yes and No, I had an old account with Istorya that hasn't been used for a long time and I kinda forgot about it that's why i signed up for a new account. As for "this", you may want to rethink your question. This "this" is not a "just" , because this "this" may be a big and serious concern to other people especially to vipvip68 and bfg9000

always an ...

[Galadriel]

Been to pod5 for a couple of times with cousins. We prefer pod5 over k1 because of the scoring system.
staff - very ok, very courteous.
system - mas daghan ka ug choices sa songs.
food - lami
drinks - varied, they have beers to vodka cruisers to name a few

basin gani ang staff na lang ang samokan namo, kay mag sige lang mi ug extend sa among time.

[edgeowns]

Oh I had one problem with scoring system. There was a time none of us sang, but the song continued playing, and it got an 87! But thats about it.

[bfg9000]

Oh I had one problem with scoring system. There was a time none of us sang, but the song continued playing, and it got an 87! But thats about it.

That's a known issue with my scoring system.

Basically, the way my scoring system works is by comparing your recording (which includes your singing) to the 2nd track on a videoke song (most videoke songs have 2 audio tracks, one instrumental, and one with vocals). It analyzes pitch, not amplitude (volume). You start with a score of 100, and points are subtracted the farther your recording's frequencies are from the 2nd track (the with-vocals track on the videoke file). The process involves a lot of math which takes a lot of CPU time, so I don't analyze the whole track, just a percentage of it (read a block of data, skip 9, read one, skip 9, etc.).

Problem is, many videoke songs' first track (which should be instrumental or minus one) contains a very faint vocal track, so with no one singing, the system picks up this faint track (which is, of course, identical to the original 2nd track with vocals). Thus it'll give a high score. I have tried to alleviate this by removing very faint sounds from the analysis, but since videoke files are sourced from different publishers, with different volume levels, different recording methods, etc. it's almost impossible to make a one-size-fits-all solution.

With more audio post-processing, this flaw can be further alleviated, however it will also result in a much longer time to compute the score. So the solution implemented is basically a compromise between time taken to compute the score, and accuracy. So it'll never be even close to being as accurate as Simon Cowell, but it's should be a lot more accurate than the scoring systems in those cheap karaoke boxes (which don't even look at pitch, just volume). And you get the ability to record your performance digitally to CD.

Note that the system's score only takes pitch and timing into account, not voice quality (which is pretty much impossible to quantify in mathematical terms). So you don't have to have a nice voice, just the ability to sing in tune, to get a good score.