Porting Miroslav Philharmonik is easy

[Otto Halmén]

First, I'm sorry for the absence. Been both busy and low on energy as of late. Lots of User Music to catch up on, and lots half-formed thoughts to spell out and post to the threads. But I'll get there.

Anyway, I made a discovery.

If you have the old, discontinued Miroslav Philharmonik lying around but don't want to use SampleTank or the built-in VSTi to play it, porting the samples manually is super-easy.

The samples, despite the .stx extension, are just plain audio files! No encryption, no secret compression, no nothing. You can use Audacity to open them and save them as .wav, although you'll have to split them manually since each instrument is merged into a single chunk of audio data.

To open them, fire up Audacity, select File -> Import -> Raw Data, and select one of the .stw files. In the next dialog, select:

• Encoding: Signed 32 bit PCM
  
• Byte order: Little-endian
  
• Channels: 1 Channel (Mono)
  
• Start offset: 0 bytes
  
• Amount to import: 100%
  
• Sample rate: 44100 Hz
  

Voila! No reverse-engineering needed.

As a side note, I find it a bit strange that they'd use 32 bit integer PCM for the final product. According to this article, the library was recorded in 20-bit and edited in 24-bit. Since even the benefits of 24-bit over 16-bit are debatable, I can't help but feel that up to half of this 7 GB library is just wasted space.

[kmlandre]

(09-26-2017, 12:54 PM)Otto Halmén Wrote: Anyway, I made a discovery.

If you have the old, discontinued Miroslav Philharmonik lying around but don't want to use SampleTank or the built-in VSTi to play it, porting the samples manually is super-easy.

The samples, despite the .stx extension, are just plain audio files! No encryption, no secret compression, no nothing. You can use Audacity to open them and save them as .wav, although you'll have to split them manually since each instrument is merged into a single chunk of audio data.

ARE YOU KIDDING ME?!?!

I spent literally *WEEKS* "resampling" just *tons* of those patches in an all-for-naught effort to convert them to Kontakt only to just walk away out of boredom!

And now some genius comes along and tries the obvious?!?!

AAAAAAAAAAAAAAARRRRRRRRRRRRRRRRRRRRRRRRRRRRRRGGGGGGGGGGGGGGGGGGGGGGGGGGHHHHHHHHHHHHHHH!!!

Also:

THANK YOU.

[bigcat1969]

Clever Otto very clever!
also welcome back.

[kneedeep]

He shoots, he scores.

[Samulis]

Winner winner, chicken dinner!

I have the ANCIENT Miroslav Vitous Symphonic Orchestral samples: String Ensembles 2.0 from 1995 for Akai format samplers on CD sitting around. Utterly stunned at the quality of the samples from then, I could actually see myself using them in a legitimate piece of music and many people not noticing the samples are freakin' 22 years old! Well, probably with a bit of fancy modwheel programming to make them crossfade and add some filters, reverb, etc.

I've noticed Audition seems to default to 32-bit in many cases, like with .ogg Vorbis files, I think it automatically lists them as 32-bit even when they're not.

Also, any chance this will work in the batch processor in Audition or do you have to do this manually? Maybe a simple rename to .wav works? Also why are you converting in Mono? I'm pretty sure the entire library was done in stereo (and that was one of the huge selling points at the time my Akai version came out over say, using default EMU sounds or something, many of which were mono).

[Otto Halmén]

(09-27-2017, 04:38 AM)Samulis Wrote: Also, any chance this will work in the batch processor in Audition or do you have to do this manually? Maybe a simple rename to .wav works? Also why are you converting in Mono? I'm pretty sure the entire library was done in stereo (and that was one of the huge selling points at the time my Akai version came out over say, using default EMU sounds or something, many of which were mono).

I'm not familiar with Audition, but look for a raw data import function. That's what you'll want to use, since these files don't have headers or metadata, just the audio data itself. Audacity also has batch processing, although I haven't used it myself yet.

I'm 100% sure all of it is mono to begin with. Press and forum posts give conflicting results, although Miroslav Philharmonik songs do sound like it's a bunch of panned mono tracks. Most importantly, the files themselves don't resolve to anything else than 44.1 kHz 32 bit integer mono PCM audio when decoded. If there ever was a stereo Miroslav Vitous Symphonic Orchestra, it's not in Miroslav Philharmonik.

Also, I personally like the default E-MU sounds.

[Mattias Westlund]

Miro 1 is not mono, so if that's what you're getting you're either missing half the data or the import sums it to mono somehow. In fact the section samples are all positional with one side clearly softer and more roomy than the other. I converted large parts of the library using (IIRC) some old version of Translator years ago and all the files are in stereo.

[Otto Halmén]

I dug deeper, and found something rather interesting.

The Miroslav Philharmonik VSTi does indeed sound like stereo when you play patches, even with all the effects turned off. Here's an ultra-short test I made: https://www.dropbox.com/s/fvw03l2lnqb6nt...t.wav?dl=0

Taking a close look at the waveforms, the right channel resembles the raw data import closely. Here, the top two are the stereo render from the VSTi, and the bottom one is the raw data import in mono:

The differences between the right channel and the raw data import look like they can be attributed to the sampler's ADSR, filters, rendering, etc. The left channel is noticeably roomier and softer than the right one, just like Mattias recalled.

Where, then, is this left channel located?

Nowhere I could find it, at least. But I had an idea.

What if the stereo is fake? What if it's created by sending the mono sample through subtly different processing for the left and right channel respectively? We've all done this in a DAW at some point, haven't we?

Therefore, if I were to replace the staccato samples with a custom, easily recognizable waveform, surely it would show up in both the left and right channels if that were the case. Audacity allows you exports raw data, so the process was easy.

I replaced the entire "Cellos Stacc.stw" with this:

I rendered the above test project with the replaced sample. Here's what I got:

Here's the sound file if you want to look at it yourself. Turn down your volume before listening: https://www.dropbox.com/s/o0ubxzomyubth5...d.wav?dl=0

This seems to indeed be the case. The impulse train shows up in both the left and right channels. It appears that the sampler adds slightly different ADSR and slightly different filtering, as well as flips the polarity of one channel.

I might test with a few more different waveforms, but the evidence is compelling.

Stereo would have been cool, though.

[Otto Halmén]

I have no idea if he'll see this, but I figured I might as well try. Looking forward to his reply.

[Samulis]

Looks like you got a bit of a reply there! Let's see what Miroslav says.

Very good detective work, btw! I might have to boot up that CD and take a peek. I could have sworn it sounded like stereo, and no one would ever waste space on one of those old samplers by duplicating channels (the fancy schmancy duplicating and phase inverting type stuff I'm pretty sure wasn't possible in any sort of real time way on most of that old hardware, though many had haas-effect-style stereo 'creators', but those seemed to bake it into the samples if I recall right).