Ah but the pinch is more in the rounding error, not the initial division.

Do you round? (1.45=1, 1.51=2) or do you just chop (1.99=1)
More advanced methods involve dithering where you spread the rounding error across the neighboring samples, but there is much debate on what the best is. Chopping is the worst by far, of course.

On an unrelated note, are you open for donations? Do you have a paypal address or something? I've spent quite a few $ on my stereo over the last couple of months, Ive even invested money in speaker cables (where the best of my microelectronics knowledge - and I am an university microelectronics graduate - cannot justify this).. something as obviously helpful as your plugin would warrant at least a token gift.

No strings attached by the way, I wont start to make demands regarding updates etc

Quote

Ah but the pinch is more in the rounding error, not the initial division.

Do you round? (1.45=1, 1.51=2) or do you just chop (1.99=1)
More advanced methods involve dithering where you spread the rounding error across the neighboring samples, but there is much debate on what the best is. Chopping is the worst by far, of course.

As it's just a simple integer calculation it will just round off, but your suggestion of upsampling to 24 bits before hand will effectively give you an 8 bit of fractional component to pass on to your DAC. It would however be relatively easy to make it round up or down correctly by adjusting the calculation as so;

v2 = (v1 * vol + 127) / 255

I'd be impressed if anyone could hear the difference though. Dithering is a whole other kettle of fish of course.

One other thing to bear in mind, I'm not entirely sure what the HQSoftProc plugin actually does to each sample. If it just multiplies the sample by 256, then this theory would be sound, as the highest possible value, 0x7fff would now be 0x7fff00. However, if it scales the sample, i.e.

v2 = v1 * 0x7fffff / 0x7fff

this will give a completely different result as 0x7fff would now be 0x7fffff. I'll have to make up some test tracks, pass them through and look at what's coming into my plugin to be sure just what it's doing. I wouldn't be surprised if it's also applying dither too.

Quote

On an unrelated note, are you open for donations? Do you have a paypal address or something? I've spent quite a few $ on my stereo over the last couple of months, Ive even invested money in speaker cables (where the best of my microelectronics knowledge - and I am an university microelectronics graduate - cannot justify this).. something as obviously helpful as your plugin would warrant at least a token gift.

No strings attached by the way, I wont start to make demands regarding updates etc

It's kind of you to offer but I'd rather not get into accepting donations for the work I've done on the plugin. I originally rewrote it for my own use as I was fed up of the original one crashing if you happened to look the wrong way at the computer, so rather than just keeping it to myself it seemed to reasonable to offer it to others, along with the source code so that anyone with suitable knowledge could adapt it as they saw fit.

Personally, I'm very happy with the plugin as it stands, in fact I was pretty happy way back on version 2.5, although I'm a lot happier with it since I rewrote it from the ground up and this probably accounts for the slowdown in development in recent months just as much as my busy life.

I now listen to all of my CD's as lossless rips from media PC instead of using my ludicrously expensive TagMcLaren DVD32FLR (which I still use for DVDs, so it's not going to waste) as I can no longer tell any difference. Without the plugin, the crimes of KMixer become glaringly obvious, so I use it for all the music I play and it never gives me any grief.

I'd still like to get to the bottom of some of the problems other people are experiencing, but without the relevant hardware it's a bit of a guessing game trying to work out what's going wrong and I've got far too many other diversions to occupy my time just at the moment.

2 puntloos: I can't use KX driver because it still not supoport AC3 passthough mode for my damn soundcard. I think I can buy new soundcar d

2. Superfluous log file creation:

How about this?

static void InitLogging(void)
{
    FILE *fh;
    fh = fopen(LOG_FILE, "rb");         // check if exists
    if (fh) {
        fclose(fh);
        fh = fopen(LOG_FILE, "wb");      // clear log
        if (fh) {
            fclose(fh);
        }
    }
}

2 puntloos: Yes i have compiled this feature.

End(force=1)
EndInternal() - mode=0 force=1
Attempting to close player thread
Gapless mode buffer underrun - Player Thread is stopping KS player
End(force=1)
EndInternal() - mode=0 force=1
Player Thread - closing
Player thread has closed
Destructed KSCore
input buffer configured to 16384 bytes
output buffer configured to 8192 bytes per packet
Enumerated Filter:
  Friendly Name:
  Sound Blaster Audigy
  Filter Type = 1


Enumerated Filter:
  Friendly Name:
  Creative Sound Blaster PCI
  Filter Type = 1


Enumerated Filter:
  Friendly Name:
  M-Audio Audiophile USB Line
  Filter Type = 1


Enumerated Filter:
  Friendly Name:
  M-Audio Audiophile USB SPDIF
  Filter Type = 1


Constructed KSCore (Input Size:16384) (Output Size 8192) (Output Cnt 8)
OpenOutput - Current Settings:
Chosen Filter:       M-Audio Audiophile USB Line
Input Buffer Size:   16384
Output Buffer Size   8192
Output Buffer Count: 8


Attempting to CreateRenderPin with the following WaveFormatExtensible (first attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 22
Samples.wValidBitsPerSample = 16
dwChannelMask               = 00000003


Failed to create pin on first attempt!

Attempting to CreateRenderPin with the following WaveFormat (2nd attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 0
Samples.wValidBitsPerSample = 0
dwChannelMask               = 00000000


OpenOutput - Current Settings:
Chosen Filter:       M-Audio Audiophile USB Line
Input Buffer Size:   16384
Output Buffer Size   8192
Output Buffer Count: 8


Attempting to CreateRenderPin with the following WaveFormatExtensible (first attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 22
Samples.wValidBitsPerSample = 16
dwChannelMask               = 00000003


Failed to create pin on first attempt!

Attempting to CreateRenderPin with the following WaveFormat (2nd attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 0
Samples.wValidBitsPerSample = 0
dwChannelMask               = 00000000


OpenOutput - Current Settings:
Chosen Filter:       M-Audio Audiophile USB Line
Input Buffer Size:   16384
Output Buffer Size   8192
Output Buffer Count: 8


Attempting to CreateRenderPin with the following WaveFormatExtensible (first attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 22
Samples.wValidBitsPerSample = 16
dwChannelMask               = 00000003


Failed to create pin on first attempt!

Attempting to CreateRenderPin with the following WaveFormat (2nd attempt):
Format.nChannels            = 2
Format.nSamplesPerSec       = 44100
Format.wBitsPerSample       = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize               = 0
Samples.wValidBitsPerSample = 0
dwChannelMask               = 00000000

Hi, where can I find the source code of version 2.5? Thanks in advance.

I'm currently using a Chaintech AV 710 card with the kernel streaming output because it has (24) bit-perfect output.

With all the Via drivers which I tried, 3.10A being the most stable, I noticed that Winamp's time/song position is wrong. A song which should play for 2 minutes for example, goes on to about 2:40, while still sounding fine.

Foobar KS and Winamp with DS work fine; is this a bug in the kernel streaming module?

int KSCore::GetKSTime(void)
{
    unsigned int bytes_per_sec, t;
    bytes_per_sec = wfx_cur.Format.nAvgBytesPerSec;

    t =  written_data / bytes_per_sec * 1000; // time to nearest second in mS
    t += (written_data % bytes_per_sec) * 1000 / bytes_per_sec;    // add in fractional component

    return t;
}

int KSCore::GetOutputTime(void)
{
    float packet_lat, out_lat = 0.0;
    unsigned int bytes_per_sec;
    int t;
    bytes_per_sec = wfx_cur.Format.nAvgBytesPerSec;
    packet_lat = (float)packet_size * 1000.0 / (float)bytes_per_sec;
    for (int n = 0; n < num_packets; n++)
    {
        if (WaitForSingleObject(events[n+EV_KSPACKET], 0) != WAIT_OBJECT_0)
        {
            out_lat += packet_lat;
        }
    }

    t = GetKSTime(); // get time written to KS interface
    t -= (int)(out_lat); // apply output latency

    return t;
}

int KSCore::GetWrittenTime(void)
{
    float in_lat;
    unsigned int bytes_per_sec;
    int t;
    bytes_per_sec = wfx_cur.Format.nAvgBytesPerSec;

    in_lat = (float)input_buffer_len * 1000.0 / (float)bytes_per_sec;
    t = GetKSTime(); // get time written to KS interface
    t += (int)(in_lat); // apply input latency

    return t;
}

What could also be a quick and dirty solution is to add an extra configuration option where a time correction can be given. For example:

A song that is 3:10 and plays for up to 4:14 can be corrected with: 190/254=0.748 as it probably is the same for all my other songs as well with this card & driver...?

Not a perfect solution by far, but maybe it is fairly easy to implement?

Attempting to CreateRenderPin with the following WaveFormatExtensible (first attempt):
Format.nChannels            = 2
Format.nSamplesPerSec      = 44100
Format.wBitsPerSample      = 16
Format.nBlockAlign          = 4
Format.nAvgBytesPerSec      = 176400
Format.cbSize              = 22
Samples.wValidBitsPerSample = 16
dwChannelMask              = 00000003


OpenOutput: Successfully Created Pin, Format - BPS: 16 FREQ: 44100 NCHANNELS: 2 VALIDBPS: 16

Hi Steve,

Doesn't work on my win vista

this have compatibility with new OS Vista?

thanks.