CONVERTING MP3 TO FLAC

Lots of audio converters will do this, but none will magically give you lossless audio from doing so, which begs the question, why? XLD, dbPoweramp, etc. will all do it, though. 

Does it actually improve the SQ of an MP3 when you convert it to FLAC by these converters?

No. You can't put back what's been removed due to the lossy compression of MP3. FLAC is a lossless compression method so when you compress and decompress a FLAC file nothing is lost. When you compress using MP3 it's a lossy process so information is discarded. 

dB PowerAmp  - easy to use - I can recommend it. It has an advantage of being a ripping software too. So you get the best of both worlds.

As Chris and James wrote - not much point converting MP3 - the information is lost forever in them. What the MP3 'creation process' does it simply removes certain parts of an audio spectrum, which the designers considered to be less important. Once it is removed, it's 'gone forever' in that MP3 file.

dbpoweramp is my recommendation. 

Foobar 2000 is a free alternative, very easy to use, can also do metadata / tag editing.

The suggestions above work, but why do you want to do this?.... it provides no benefit just additional significant overhead, and depending on the quality of your MP3 decoder in your streamer versus software could actually provide a worse SQ experience.

Although you can never replace the bits that are lost by converting to MP3, some people have reported that converting this lossy file back to FLAC or WAV still produces an improved, or at least changed sound. There is nothing to lose by trying this with a few albums to see for yorself (apart from soace on your hard drive) although personally I have never bothered to try it. 

I have converted mp3 320k (and som of my WAV files as well) to FLAC to be able to use replay gain tags to reduce the level.  

Minimserver is able to use the replay gain tags added to the FLAC files to reduce the level in the WAV stream to the ND5, when transcoding on the fly. I find this very useful with a lot of music, where the recording level is much to high, the added replay gain is often close to -10 dB. 

As far as I know the Naim streamers can't use the replay gain tags set directly on the WAV files... (?)

But apart from that I can't see a reason for converting MP3 to FLAC thus making the files some 3x larger (my experience). 

Claus

Claus - however changing gain this way is not going to be good for aliasing unless noise (dither) is artificially added. Definitely avoid at 16 bit - 24 bit less of an issue perhaps... also one has lost bit perfect replay by doing this.

For PCM (for optimum SQ) it is best to leave the sample sizes well alone and change the gain separately for optimum performance and obviously for bit perfect reproductions and reconstruction

Simon, I have much less insight in these matters than you, and probably most other good people here, but my understanding is this:

I have replay gain added as tag info, this should not influence the music content of the files, and I can always remove the tag info again, or skip it during replay. 

During transcoding I have the signal "upsampled" - I expect that this way the detrimental effect of the replay gain adjustment would be minimized.

I hope that this method doesn't compromise the SQ to any significant degree - ??

To me quite a few tracks/albums were not a good listen because they had much too high output level. I know that replay gain can't cure the issue of limited dynamics on these. But it does, for me, make it much easier to listen to these. To name just one this could be Lana del Rey: Born to Die. This album, and others, is almost impossible to play without RG at low volume on my ND5/nDac/52/250/SBL system. 

It's also (for me) a huge benefit that it reduces the need for adjusting the volume on the preamp between the tracks. 

I guess that without RG tags I would probably completely avoid playing as much as 5-10% of my cd collection. And avoiding major differences in sound level from track to track is really convenient, and appreciated by the wife as well 

Claus

I used to get an app or uniti 'hang' when swapping between music of different 'filetype', this is going back some years btw. I cured it in my home by making all files 'the same' (it never hung swapping from one flac file to another) so I chose flac over wav (metadata, filesize convenience) after deciding any perceived sq differences were negligible. It should be noted any possible problem with my system/devices may have been cured long ago by various updates. I now have very little to do with aac or mp3 but when I have a lossy file I convert without hesitation. I keep backups in the original form I received them (just in case I go up a format blind alley!)

And to answer the OP: dbPoweramp.

Claus posted:

Simon, I have much less insight in these matters than you, and probably most other good people here, but my understanding is this:

I have replay gain added as tag info, this should not influence the music content of the files, and I can always remove the tag info again, or skip it during replay. 

During transcoding I have the signal "upsampled" - I expect that this way the detrimental effect of the replay gain adjustment would be minimized.

I hope that this method doesn't compromise the SQ to any significant degree - ??

To me quite a few tracks/albums were not a good listen because they had much too high output level. I know that replay gain can't cure the issue of limited dynamics on these. But it does, for me, make it much easier to listen to these. To name just one this could be Lana del Rey: Born to Die. This album, and others, is almost impossible to play without RG at low volume on my ND5/nDac/52/250/SBL system. 

It's also (for me) a huge benefit that it reduces the need for adjusting the volume on the preamp between the tracks. 

I guess that without RG tags I would probably completely avoid playing as much as 5-10% of my cd collection. And avoiding major differences in sound level from track to track is really convenient, and appreciated by the wife as well 

Claus

Hi Clause, well first if you are happy with the resultant sound then that is the main thing, and the fact you may be comprising the sample data with your gain settings is neither here nor there...

However if the gain is being adjusted at the digital level, the sample values have to be reduced or increased.... and as we are dealing in integers and not floating value sample values, errors will occur, and digital noise will be added.

Upsampling doesn’t affect this ... that adds another dimension of mathematical approximation errors... so you start to get the idea, if you want the sound to be as close as to what it was when it was mastered, best not fiddle, and let the gain be adjusted in the analogue domain

Thanks Simon, I get your point, I think. 

To ensure the highest possible SQ, as in the most original, I should not do anything but rip in wav and play the files as they are. 

To me, with some albums, the advantage of reducing the level with a replay gain tag is more important than a slight perhaps hardly noticeable drop in sq. I did a few A-B tests with the same tracks with and without RG, and with some I prefer the RG version, with others the opposite, and with some I can't say for sure  

When I prefer the RG version, the gain is usually approx -10 dB. So I usually only apply RG tags if the value is above 7. 

I could of course find an attenuated in from nDac to preamp. That would reduce the level making it easier to adjust volume at very low levels. But it wouldn't even out the sometimes huge difference in sound level from track to track. Which also is a reason for me to use RG tags  

Clais