Emulating older CDP with MPEG Audio Decoder

[giro]

*Purely Experimental* -

 

Back to basics - I'm after a music player that is strictly fixed point front-to-back in order to emulate the process Philips employed for getting 16bit performance out of a 14bit mulitbit.

However instead of 14bit, I want to experiment with 8bit (which is not far from vinyl hifi performance regarding SNR).

Essentially I want to emulate vinyl.

 

I did research and discovered fixed point decoder MPEG audio decoder (MAD)

 

CoolPlayer is an option but no re-sampling or dither plugins are available for it.

I'm sure it wouldn't be too hard to impliment sox plugin for it eitherway.

 

I also found MAD plugin for winamp, I'm sure to achieve "8bit", the plugin upsamples x64 and dithers down for clean 8bit out - though it would be nice to know what it's going on internally and change resampler to sox (also mad based).

 

 

As I say, the mad plugin for winamp probably is my best bet but incorporating sox would top it off nicely - i'm contacting the developer for more info.

 

Unsure if winamp is fixed point front to back though ~ ?

 

 

Any thoughts?

 

 

Cheers

[iago]

I'm not quite sure where your journey is leading you, but kindly allow me to question a few of your assumptions.

 

If you are referring to the first generation of Phillips DAC chips, these were not multi-bit. They used R-2R networks and were simply limited to 14 bit resolution.

 

I think you are aming too low with the 8-bit resolution for Vinyl. This would give you a SNR of 48 dB. I would assume at least 60-70 dB noise floor for a standard turntable assembly, which would give you around 11 bit resolution.

 

As far as I know, Philips was decoding the CD in a normal way (I think you have no choice, because of the necessary format translation from Reed Solomon to LPCM). Only the final D/A conversion was 14 bit. To overcome this limitation they used 4x oversampling, which gained them 1 bit in resolution.

 

Ideally you should create your own plugin for WinAmp, first upsampling the original data and afterwards decimating the resolution. You could also do it on file level; upsampling is fairly easy, but I'm not sure if the available software allows you to arbitrarily manipulate the resolution (there may be hardware or file format restrictions also).

[Skeptic]

I'm not sure why you would want this, but JRiver has a bitdepth simulator you can enable in the Parametric EQ DSP.

This can be set to any bit-depth you want from 4 to 24 bit, with dithering being optional.

[giro]

Ok, thanks for the response. y