Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[thecrow]

I have a Tom Petty album in ape format that is encoded in HDCD.
When analysed in MC it is correctly identified as HDCD.
When converted to flac using MC and re-analysed it loses the HDCD identification.
When converted to flac using Xiph's command line encoder and re-analysed it retains the HDCD identification.

Upon further investigation I found converting to any lossless format using MC removes the HDCD signature, whereas converting with other programs does not.

Is there a problem with MC's format converter that means it is not lossless?

[Hendrik]

My guess is that MC actually performs the HDCD decoding when you convert the audio.

[thecrow]

After conversion the file is still 16bit.
Comparing the audio analysis of the two files they show identical values for peak level, dynamic range, etc 

[dtc]

Are you applying any DSP during the conversion?

[thecrow]

No I believe I am just doing a simple format convert between lossless formats with no DSP.
If you have any HDCD files I would be interested if you see the same when converting format.
On the original file using the tag editor check the HDCD tag has a value of 1.
Convert the file to another lossless format such as wav, ape, or flac.
Then recheck the HDCD tag, for me this is now showing 0 on the converted file.

[Hendrik]

I still think its simply decoding the HDCD data. The bitdepth is determined before any decoding/encoding takes places, so it would not change the bitdepth  on account of HDCD decoding unless you tell it to.

[dtc]

I just converted a flac file to ape and back to flac.  I  set HDCD to -1  in both files and did analyze audio. In both cases it recognized the new files as HDCD and set the flag to 1. So, it seems like it is doing the right thing for me.  I made sure the DSP check box was not checked.

[dtc]

I still think its simply decoding the HDCD data. The bitdepth is determined before any decoding/encoding takes places, so it would not change the bitdepth  on account of HDCD decoding unless you tell it to.

If it were to do that, it would be a mistake in my opinion. Some people still have HDCD DACs and do not want the software decoding to happen. Forcing the decoding without an option to disable the decoding would mean that MC could not reliably be used for converting HDCD files from one format to another in a bit perfect form.

There was a discussion early on about using the convert function to create decoded 24 bit files. I think that is a good idea, as some people may want the decoded files to play outside of MC. However, it would need to be an option, for the reason stately above.


You can check to see if the new file is 24 bits simply by checking on the file size of the file after conversion.

If MC did a decode during a conversion, once converted, the new file would be a separate 24 bit file and the bitdepth of that should be reported as 24 bit.  If, in fact, MC created a new file that was 24 bits but did not change the bitdepth tag to 24, then I believe that would be an error. 

[thecrow]

I just converted a flac file to ape and back to flac.  I  set HDCD to -1  in both files and did analyze audio. In both cases it recognized the new files as HDCD and set the flag to 1. So, it seems like it is doing the right thing for me.  I made sure the DSP check box was not checked.

Thanks for that, it spurred me on to look deeper as to what I was doing that was breaking this.
And it is the fact that I have TPDF dither selected in Advanced Audio Options.
With dither set to None or JRiver bit-exact then I retain the HDCD identifier with TPDF I lose it.

So when converting between lossless formats with TPDF selected it is changing the audio by re-dithering and so is not a bit for bit conversion.

[dtc]

Good detective work.  I am a little surprised that dithering is applied in a conversion between lossless formats. I would have thought it was just unpacking and repacking  the bits unless DSP is checked.

[Awesome Donkey]

TPDF shouldn't kick in unless you're dithering away from the source bit-depth (16-bit in this case), right? If it's using TPDF dithering without changing bit-depth when converting audio, that kinda sounds like a bug to me. At least, that's what makes sense to me, not using dithering when converting audio if the bit-depth is going to be the same and ONLY use dithering when there's a change in bit-depth with the conversion. But yeah, IMO it definitely shouldn't be dithering/resampling when doing a lossless to lossless conversion at the same bit-depth/sample rate.

Seems to me that's not exactly a 'lossless' conversion in that case if TPDF dithering is being applied when there's no change in bit-depth being made in the conversion.

[thecrow]

I think re-dithering would be valid after any DSP was applied even if the bit depth was unchanged.
But I do think it is wrong when simply doing a file format conversion that is not changing the audio in anyway.
So I would agree it looks like a bug.

[dtc]

Agreed.

[lepa]

There is bit depth setting in the encoder options. Was it set to automatic which you would think would be using the same depth as source? There isn't "same as source" option available

[thecrow]

My library is a mix of 16bit, 24bit and 1bit files so I always leave bit depth set to automatic, as it was in this case too.

[dtc]

I had mine set to 16 bit in Convert with JRiver Bit Exact dithering set.