Stereo Tool

Vamprecords: Is that also the case when you turn the Loudness slider down to - say - 1.01 or so? If so, it's a bug, if not it's just the sound that you get when setting the Loudness slider too high. The higher you set the slider, the worse the sound gets. (Values above 2.5 really sound horrible).

Tekie: In an ideal world, you would be right. Unfortunately, if all the other stations are using huge (and I mean HUGE) amounts of limiting, stations that don't sound really soft (although clearly better quality wise). Unfortunately in many places there's a "loudness war" going on where every station tries to sound louder than the others. Indeed at the cost of quality and vibrance and often causing very unclear high frequencies. I'm trying to get my Loudness filter to sound as naturally as possible - of course when you use high values to increase the volume by - say - 12 dB, that's not really possible.

I live in the Netherlands. If I turn on a German radio station in my car I hear almost nothing because German radio stations broadcast about 3-4 times softer than Dutch stations... So if a station doesn't do anything at all to reach a similar volume to other Dutch stations, many people will think that the station is broken or broadcasting silence and just skip it... On the other hand, when the reception is good those German stations sound almost as good as a CD...

About the bad sound quality: Is that what you are getting with (the latest version of) Stereo Tool or is that a general remark about limiters? (And if it is Stereo Tool, is it worse than others that you have used?) If you turn ++ on and keep the Loudness slider low (say around 1.5 or so), there should be very little effect on the sound quality.

I've just finished comparing my own filter with some others, and I found that while other filters have more distortion, mine has less bass and slightly more pumping when loud high frequencies are present. I have a solution for the bass sounds (which is already used when ++ is turned off) that will make them sound louder without increasing the amplitude - that should also give more headroom to the higher frequencies and make the entire thing sound better. I'll look into that when I have some time left.

Note: Make sure that the Final Pre Limiter and Final Limiter are turned ON. That wasn't needed in previous versions, but the latest version sounds a lot better with those two turned on.

Hans, that was a general remark.

I'm aware of the loudness wars. Although, I believe their efforts are in vain.
Where is the end to it?

I believe if one was to concentrate on the quality of the sound and not so much the
loudness (as in the loudness wars) you would end up with a "decent sounding" station.

Thats what I aim for.

To trade off quality for loudness, I believe, is going down the wrong path.

About your loudness filter ... It sounds good, so long as one doesn't turn it up too far.
1.25 would be as far as I would go with it.

I am the same way with mastering limiters or multi-band maximizers.
Just enough to give it a slight boost.

What can I say? I'm moderate on "loudness" when it comes to audio.

________________________________________________________

Another point I want to make:

There are some who think down mixing - rotation - seperation of stereo channels
into two mono channels, then remixing the whole kettle of brew into something
LOUD and FAT at the end of 20 to 30 stages is "audio engineering".

I left out alot in the above statement .. but, you get the picture.

By the time they are finished with contorting the audio to meet the
sound they are looking for ... the original artist material is lost
into a mire of squashing, pumping, expanding, harmonic duplication, etc.

I don't know .. maybe i'm being too critical.

cheers!

Hi Hans,

I tested again, and now with the ++ set to 1 and still the sound is really bad.. it's not really a problem for me but i was just testing the tool..

For your info i'm using a M-Audio audiophile 192 sound card.

Martijn

Vamprecords: Ok, then it must be the bug that I mentioned earlier. I don't know why it happens always on your system (which Stereo Tool flavor are you using? Winamp, VST, stand alone? SSE2 or non-SSE2?) I will have to look into this.

I'm using the winamp plugin in SamBroadcaster.

Tekie: Based on your comments I went back and checked the sound quality that I'm getting against the theoretical optimal quality. The theoretical optimal quality was better (cleaner, punchier, more and deeper bass). Tonight I've been playing with some settings, and I think I have now found a much easier way of reaching the sound quality that I'm aiming for. And at least my first results seem to indicate that this new approach will sound much better.

I'll try to get something working in a few days.

Vamprecords: That new much simpler version should also fix your "bad MP3-like" sound problem, whatever is causing it. The new approach is so much easier (probably about 4 or 5 filters instead of the current 30 (!) filters) that I can't do too much wrong

Also, if this works, it will greatly reduce the latency and CPU load (after all, 4-5 filters instead of 30 makes quite a difference).

Ok, here's an update!

I have replaced the Loudness without ++ filter. It reaches approximately the same volume as the ++ version, but with:
- MUCH cleaner sound
- MORE deep bass
- Much more punchy
- GREATLY reduced CPU load

Some notoriously difficult sounds might have become slightly more distorted. But the majority of the sound has become far less distorted. I might still need to tweak it a bit, but this is by far the best version to date.

You can download the new versions here:
Winamp DSP plugin: http://www.stereotool.com/download/dsp_ ... _BETA2.exe
Stand alone version: http://www.stereotool.com/download/ster ... _BETA2.exe
VST plugin: http://www.stereotool.com/download/vst_ ... _BETA2.exe
VST plugin, no SSE2: http://www.stereotool.com/download/vst_ ... _BETA2.exe

When I'm finished tweaking it I'll replace the normal beta (instead of using this BETA2). I will also remove the ++ option as it is no longer needed.

Like yu I am using ST for both SAM version SAM BC and SAM DJ.

Are you streaming direct or using kernal streaming for SAM.
I am making the assumption that you are using 4.36 and IF your sound card allows it then using the kernal streaming for output with/on SAM will produce a much better sound quality.

Note not all sounds cards let you so you have to experiment.

I use the kernal to a great success and therby allow me to also utilize feature setting of/from my sound card in addition in particular EAX, EQ, 3D.

Iam not sure IF Hans has ever tested the difference between using direct output streaming with kernal streaming IF NOT then it may be worthwhile doing by virtue of the vast difference between them.

regards

mapleleaf

This is a great version, I like it very much my avg, cpu load is 28.42 - 33.21 %

I've done several things to help reduce the cpu load
1. deactivated indexing service on my system << not really needed
2. deactivated restore service << only activated when I need to perform a restore.

both of these will place overhead on the cpu during or while SAM and ST are running.

I did notice that a 10% cpu load will occur when one selects HD.

Btw, did you ever check out that tcp/ip issue with the 'time-wait status' ?

regards

mapleleaf

Hi Mapleleaf,

> I did notice that a 10% cpu load will occur when one selects HD.

What do you mean by HD?

> Btw, did you ever check out that tcp/ip issue with the 'time-wait status' ?

Nope, not yet anyway...

About kernel streaming: Kernel streaming bypasses the Windows kMixer, an annoying piece of software that combines audio from different programs into one audio stream that is sent to the sound card. Especially older sound cards needed this because they didn't have the capability to play multiple streams simultaneously.

Because there's only one stream, kMixer often has to downsample or upsample streams before they can be combined. Often the drivers can be configured to always use a specific sample rate and number of bits (usually by default 44.1 kHz, 16 bits). Anything sent to the card with another sample rate is converted in that case. The kMixer sample rate converter is really bad, which often leads to really bad sound quality when this occurs.

I'm using the ESI Juli@ card for MPX output. If ANYTHING happens to the MPX signal, it won't be received correctly by the radio anymore. Since it is received correctly, I know that my sound card perfectly reproduces whatever I send to it. So there's no need to use kernel streaming. The driver switches automatically (which only works if you only use one program at a time), so kernel streaming won't give better results. On Windows Vista however, the driver doesn't support automatic switching and I do have to configure the intended sample rate and bit depth manually. (Without doing that I also cannot broadcast the MPX signal).