Converting & Upsampling music files mp3, AAC, 256kbs, 320kbs

[John Dyson]

You want to save the recording in exactly the same form that you received it for your own archival/backup purposes.

 

If you want to play on an app that doesn't support the format, then do a conversion from the ORIGINAL file, and, if possible, use a format with a quality/bitrate combination that is theoretically better than the original format.

 

If wanting to use a lossy format, and need to convert, then use the best bitrate/quality of that lossy format:   if I must, I use mp3/320k, opus/384k (but 256k is good enough), aac/256k or better.   When you swap formats, sometimes you end up the worst of both -- that is the reason on the NEW copy, use the highest quality of the lossy format.   As a tradeoff, where you want to save a little space, then mp3/256k, opus/256k, aac/256k.   192k is a good second level for high quality (mp3 might be a little worse -- but nothing to worry about.)  No format is 'perfect' at 128k, but opus can sometimes do a great, but not high fidelity  job at 96k -- I have heard 64-80k (I forget which one)  and it is suprisingly good, but not high fidelity.   Don't convert down below 192k if you want to keep great copies of your material. 192k is on the 'edge' for a long term copy.

 

If you want to archive your original at essentially full quality, willing to spend space, but don't want to waste it, then a lossless 16bit format like flac at the SAME rate or higher is great.   Depending on your software 44.1k -> 48k conversion to a flac might cause a slight loss, I'd just keep the native rate or substantially higher.   If you want to be 'careful' and keep with maximum resolution, then 24bit flac or equivalent gives some wiggle room.   Mostly, those extra bits will record the errors more accurately -- but once in a long while, might help.

 

Once a sampling has been done, the only thing that a sample rate increase does is to make certain kinds of processing easier to do.  Frankly, if I started with lossy, I'd always keep the original lossy file, and do conversions as needed (with the above guidance.)   If I really needed to downconvert to 96k for a certain device, I wouldn't be embarassed -- but ALWAYS keep your original highest quality copy.   Every additional copy layer, where a layer of lossy conversion is done, will cause SOME loss.   After the first conversion to lossless 16, there will be no loss until you convert back to lossy.   When converting to 24bit lossless, you'll forever have a pristine copy of the original lossy  file, but in a lossless form.  (Of course, this all assumes that the software is as good as it should be.)   It is midly questionable that a first conversion from certain lossy formats to 16bit lossless might cause a low, bit-level kind of error,beyond that -- all in lossless format, not changing numbers of bits/sample rate -- will be very clean.

 

John

 

[John Dyson]

On 10/24/2020 at 1:59 AM, AnotherSpin said:

 

There is no reason to convert lossy format file into FLAC it will not turn the file into lossless. If you optained file in lossy, keep it in lossy. If you care about lossless, get it in lossless.

The reason is for consistency of an archive.   The single conversion to flac will do no damage (well, maybe bit level at 16bits), but after that the  entire achive can always be flac.   I don't like messy 'is this mp3 or opus or aac or flac' or whatever.   What about the crazy m4t or whatever formats also?  Resolve it all to one format, at least IMO.

 

The negative of keeping everything in flac is space.

The positive of keeping everything in flac is everything is the same.

 

If you prefer .wav, that is the same as .flac, but takes more space.  Also have the option of 32 bit or FP, but who cares about those?

 

Additionally, no real additional loss.

Certainly, converting to a lossless format will not make the file lossless, but storage space tends to be cheep cheep cheep (or is it cheap :-)).

 

John

 

[John Dyson]

1 minute ago, AnotherSpin said:

 

The uniformity of an archive can be attractive in certain cases, there is no doubt. However, the day will come when the question arises whether certain files have been obtained in lossless initially or up converted from lossy? It is possible to find out, with the help of certain additional procedures. But do we seem to be talking about simplicity and convenience here?

I am only talking about simplicity, uniformity and convenience -- no technical merit other than an almost 100% guarantee of no new loss -- esp if choosing 24 bit or 16 bit depending on quality, while archiving the material in a consistent format.   Theoretically the higher than 16 bit capable resolution lossy formats MIGHT lose something when converting to 16bit flac, but probably much less than any conversion to another lossy format.

 

Trying to move to a different lossy format for consistency will cause loss.

 

If starting with a lossy form, it might be nice to log the original format into the flac (or otherwise) metadata.

Keeping important information in the per-file metadata is also a nice standard practice.

 

Knowing what I know now, I would have been much better organized in creating my library archive...

1)  Consider a single format (or a limited set of formats)

2)  Log origination information into the metadata, including CD ID, etc.

3)  Log original data form.

4)  Log aquisition date.

 

Trying to keep the data in parallel is frought with troubles.  if it is all in the same .wav, .flac, etc file, then

life is easier later on.

John