The filter processor is controlled by some settings in the InguzDSP.exe.config file.
- On Windows, this is in .
- On Debian/Ubuntu linux, this is in .
- On Red Hat linux, this is in .
You can open this file with a text editor and change settings. Be careful to keep the XML tags matched!
The default configuration file looks like this:
Each of these three key values has a different effect. Additionally, there are other settings available (documented below) which you can add to this configuration file.
Changes take effect on the next track played (not during the same track).
Gain (in decibels)
The setting tells the filter processor how much to increase or decrease the signal strength during processing. The appropriate value for this will depend on your particular room-correction and other equalization settings.
The default value -16 seems to be right for the majority of systems.
If you see clipping in the logfile, then consider reducing this. You should usually change the value in small increments: go to -17, -18 or -19, for example.
On the other hand, if you feel that the processor makes music too quiet in general, then check the “peak” values in the logfile, and increase the gain as you see fit (to -6, 0, or even positive values). If you increase gain too much, you will hit clipping; since clipping sounds much nastier than being too quiet, it’s worthwhile to be careful :-)
Partitions
The setting changes the way the filter processing is performed. It has no effect on the sound, but might affect performance slightly.
For most situations, this should be set to 2. You can try values of 0 or 4, to see whether there’s a noticeable effect on performance. In most cases the difference is minimal.
Dither
The setting changes the way in which the signal is dithered before sending to the Squeezebox or Transporter. Since dither is applied only to the 24-bit signal, its effect is virtually inaudible.
The various dither values are:
- 0: no dither. Values are rounded at the 24th bit. With no dither, you may be able to detect slight non-musical “edges” on quiet sounds, transients and decays.
- 1: triangular dither. This is the “theoretically correct” dither.
- 2: Noise-shaped dither (per Wannamaker, JAES vol 40, 7/8, 1992). This is only suitable for 44.1 kHz sampling rates, so even if it’s distinguishable from triangular, it’s not preferable for a general-purpose system.
- 3: Another noise-shaped dither; again, suitable for 44.1kHz sampling rates only.
Maximum filter length
The setting limits the length of the room-correction impulse. If the current room-correction filter is longer than this, it is shortened and windowed with a Blackman-Harris window. The default value is 65536. You might want to set this to a higher value if you want extremely long room-correction filters (and don’t mind using extra CPU power to process them), or set to a lower value if you find InguzDSP uses too much processing power in normal usage.
Filter resampling program
The setting overrides the program used to resample the room-correction impulse when playing music that has a different sampling rate than the impulse file. The default (or when this setting is not present) is
The setting overrides the command-line parameters provided to the resampleExe program (see above). The default (or when this setting is not present) is
where {0} represents the input filename, {1} represents the desired output sample rate, and {2} represents the output filename.
If you prefer to use a different resampler than sox (e.g. ssrc or other command-line resampler), you can change this pair of commands appropriately.
Debug
If the setting is “true”, this enables some extra debug messages in the log file.
