Memecode

Converting audio samples to dB and back

Thu, 26 Apr 2012 23:28:43 +1000

I've been writing a tool to normalize lots of audio files at once, as well as convert between various loss-less formats (particularly FLAC and WAV). In doing that I needed a way of converting between the raw audio sample maximum and dB. So I present to you my C functions for doing so:

double LinearToDb(int32 linear, int bitDepth)
{
    uint32 MaxLinear = (1 << (bitDepth - 1)) - 1;
    uint32 ab = linear >= 0 ? linear : -linear;
    return log10((double)ab / MaxLinear) * 20.0;
}

int32 DbToLinear(double dB, int bitDepth)
{
    uint32 MaxLinear = (1 << (bitDepth - 1)) - 1;
    double d = pow(10, dB / 20);
    return d * MaxLinear;
}

Another code snippit for Google to index.

XCode: error from debugger: the program being debugged is not being run.

Sun, 15 Apr 2012 11:55:57 +1000

If you are getting this in your XCode run log:

Running...
No executable file specified.
Use the "file" or "exec-file" command.
No executable file specified.
Use the "file" or "exec-file" command.
The program being debugged is not being run.
The program being debugged is not being run.

After copying a project and renaming everything... then you missed the "executable name" on the target. Click your Target, and "Get Info", then click the Properties tab, and rename the Executable to the same name as the Product Name in the Build settings.

DeleteFile failure on long paths.

Fri, 09 Mar 2012 12:10:42 +1100

Symptom: DeleteFile fails with ERROR_PATH_NOT_FOUND (3) when passed a path with a length greater than 260 characters.

This manifested for me when I copied a Windows XP home folder to a backup drive. The "Temporary Internet File" folder contains a lot of files with very long names. When you put those in a sub-folder with a long name, the total path length of those files tips over the 260 character limit. At that point Windows Explorer just fails to do anything useful on those files. In my case I just want to delete them. So I was poking around with i.File in an attempt to work out why these files can't be deleted. Turns out DeleteFile simply fails with super long paths.

The way around this is to share the drive with the long paths on it. Then map a drive to a deep sub-folder to reduce the file's path length to under 260 characters. THEN delete it using traditional methods.

Mac OS X Software Update fails to update installed application.

Tue, 21 Feb 2012 12:02:00 +1100

I've just managed to "fix" an issue that I was seeing on 10.6 where Software Update would not update Logic Express. According to Software Update, Logic wasn't even installed, however the app was there in /Applications and would run fine.

It seems that getting Software Update to re-scan that app is as simple as renaming the app. I changed the name of Logic Express from "Logic Express 9.0.1" to just "Logic Express", ran Software Update again and the latest release for Logic magically appeared. Hmmm.

So the next time Software Update is ignoring your installed apps you know what to do!

iconv v1.9.1 win32/win64

Sat, 12 Nov 2011 17:24:28 +1100

I've posted a source and binaries zip of iconv for win32 and win64 on the Libraries page.

Axefx Foot Controller

Wed, 22 Jun 2011 09:59:43 +1000

After some months of work I've finally got my Axefx foot controller kit up for sale. I've updated the index page to have both software and hardware sections... because apparently I do hardware too now :)

Wrapping Up Controller Testing

Mon, 20 Jun 2011 21:40:55 +1000

So in the last few days I've finished the controller test board, ironed out a few bugs in the firmware code, hooked up the midi ports and tested that they are sending and receiving the right data. The board looks like it's doing all the things it's meant too. Which means I can now start moving ahead on making the kit available. What's left to do is write up the documentation of how to build a kit from the parts and PCB and work out the business details, like shipping weights/costs and a purchase page.

Today I also printed out the new graphics that are laid out for the Hammond box and checked them for accuracy.

It's sitting next to the prototype box, which is a different aspect ratio. The plan is to get someone to print it out onto adhesive backed vinyl and then stick it straight onto the aluminium, cut the holes out with a hobby knife and then poke all the pots, switches and LED's through. There are some more knobs coming from the UK for both my own controller and the kits.

Controller Update

Fri, 17 Jun 2011 10:30:38 +1000

Last night I got some time to work on the MIDI controller testing. The problem I was left with is that the parameter knobs were not working, so I started with printing out the current value for param0 on the LCD and it was always 1023 (the maximum value) no matter what voltage was supplied on the input pin. So I teared down all the setup code and checked that I was doing everything right in the software... which after an hour or so of re-reading all the docs... yes; I'm doing everything right. Ok that leaves hardware. So I measured the voltage on the input pin again... yup that's right too. *sigh* Then I notice there is a "VREF" pot on the AVR stamp... wonder what that does? Tweak tweak... oooh look the value on the LCD is finally changing! Seems that VREF pot on the stamp is controlling something to do with the way the AVR reads values on it's analog in pins, so I'll have to write that into the documentation to set correctly when assembling the kit. So hurrah the parameters are all working pretty much correctly and I can test the new menu code on the hardware. I'm getting close to being done with the testing. All that remains is hooking up the MIDI sockets and checking the right messages get sent and received.

Midi Foot-controller Update

Wed, 15 Jun 2011 10:55:08 +1000

Last night I investigated whether I could compile and flash the ROM from Windows7 64bit. The compiling side was fine, I could build the binary but the program I was using to flash that binary onto the AVR chip via the parallel port doesn't work on Windows7-64bit. Which isn't to say you can't do AVR development on Win7 but it just means that different software is needed. On XP I've been using PonyProg2000 with a lot of success.

On the hardware front, I wired up all the pots to the test harness and checked they were applying the right voltage to the analog pins. I've soldered those onto the PCB traces and not the pads because it's a pain removing wires from pads and then trying to get the solder out and put something else in.

Software wise I have re-written the menu and not everything is hooked up at the moment. So I spent some time uncommented some sections of the controller code and got them compiling with the new menu layout. The structure of things in memory (both RAM and EEPROM) has changed to make things more flexible and those changes need to be propagated out to all the rest of the code. (Parts that didn't compile originally have been commented out). Also I put together a Visual Studio 2005 project for the desktop version of the controller.

And I'm loving zip ties at the moment... they make things so much more organized.

64bit Adventures

Sat, 11 Jun 2011 16:10:42 +1000

I've been playing around with 64bit builds of Scribe on windows. And it's been er... interesting to say the least.

By default the Visual Studio 2005 installed doesn't install the x64 compiler and libraries. Nice...
When you do get that installed, and then create an x64 profile in your solution you get the option to copy the settings from the 32bit profile. Which for the most part works, but when you try and compile said x64 profile, you discover this delightful error:
```
fatal error LNK1112: module machine type 'x64' conflicts with target machine type 'X86'
```
Which in a nutshell means you are linking 64bit objects into a 32bit binary. However the standard response that people get on the forums is that you have to set your Linker- >Advanced->TargetMachine correctly. And you know what? Well it WAS set correctly to MachineX64. So what gives? Well if you click on Linker->CommandLine and look in "Additional Options" you'll see /MACHINE:I386. Sigh.
When running Windows7 64bit and you put additional 32bit DLL's in C:\Windows\System32 you'd think... that since there is a "32" in that folder name that is where they go. Well no... the 64bit DLL's go in System32 and the 32bit DLL's go in "System". Gaaaaahhhhh.

But at the end of the day, there it is... 64bit Scribe: