Hello..

As stated in previous posts.. I run lots of `tests' on female vocal CDs and have been encoding them with MPC lately at the setting of -q6 per recommendations in this forum. Playing back singers such as Mary Chapin Carpenter (last night) and Loreena McKennitt (just a few minures ago) reveals a somewhat harsh, gritty, even slightly `under water' sound in a number of the passages where their voices range up in prominence (am using the --xlevel switch).

One section of M.C. Carpenter fell apart all together when accompanied by a rhythm guitar with lots of reverb. Her vocals seemed to `melt' into a pile. Have compared these mpc's with the wav files although not A-B comparisons. This is rather disappointing because the female vocals are what I like best.. and this type of `vocals' material is know to quickly show flaws in speakers and hardware.. such as back in the analog days.

What is wrong? Is there an inherent flaw? Should I go to -q7 or 8? What about another codec? After encoding, I heard a section of Sade (pronounced Shar Day) that sounded like high-IM distortion using VBR MP3 a week or two back. Went to MPC after this. What causes harshness as such in an ordinarily pure female singing voice?



Cheers..!!

i would recommend abx'ing the clips first, but if you are sure its a problem, you should upload some short clips in lossless format (flac preferably) somewhere and link to it so others can confirm the issues. no codec is perfect, but flaws should be very rare with mpc.

By ABX'ing .. do you mean A-B comparison? Isn't there an application for this? Also.. if I could edit and isolate these segments.. then you could also evaluate.

Are you using the Winamp based decoder to listen to your files? To my knowledge, older versions had a quality bug in them (I think v0.97g fixes this...?) If you were using Winamp, try mppdec --prev and listen to the files again. If not, i don't know what the problem is. It could just be the bitrate is to low - try -q7 / 8 and try again.

It's 0.98 in_mpc.dll plugin for Winamp 2.81

Will dec --prev and burn at higher q if necessary.. thanks



somehow gut-feeling I think a higher bitrate is needed for these ladies' voices..

Of course.
Try KikeG's WinABX

For other programs and all ABXing needs go to pcabx.com

Went ahead and did some A-B comparisons like your all suggested.. and didn't hear the `gritting' or distortion that I heard before (about 1:00 AM). It now seems more likely a Gain issue and maybe even clipping. This kind of courseness is the charateristic of `squareing off' of sine waves.

The settings in the Winamp MPC plugin were down to the left.. and I moved them higher. Need to better understand the GAIN and offset thing.. wish there were more DOCs or `howto' pages that clearly showed charts in dB and the plugin ranges mapped over top.


So.. let's not blame it on the codec failing so fast here..!!



p.s. Please post (aGaiN) the link for the ReplayGain discussion.

just use replaygain --auto on your mpc's, and then set the mpc plugin to album based, clipprotect off; and set the headroom to k-14 (all this is in the sticky). There will be no clipping.

Also disable the winamp eq if you are using it.

I'm sorry to see -- this post has gone far afield..

The song in question.. that sounded distorted in some of the more pronounced vocals sections.. was reduced by -8.35 dB after running the folder of output mpc files through ReplayGain. The entire album was reduced by -7.33 dB.


The output folder is [now] named in a self-explanitory way: \mpc -q6.2 -auto

Also the ReplayGain mpc plugin 0.98 settings have been changed.


The `gritty' sound is no longer audible.. I believe it's safe to say.

Sorry, the trolling should have been catched earlier.
Anyway, it's now moved to the off-topic forum.