Skip to main content

Topic: Which is the best lossless codec? (Read 353201 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.
Which is the best lossless codec?
Reply #300
(And to avoid further edition of the previous post)

Hence my considering WavPack as the best of both worlds when it comes to lossless encoding (and just to stick to the now-ancient OP).
Listen to the music, not the media.

Best,
Nilson

  • marc2003
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #301
That you only have to worry about tagging the files in a supposed lossless/lossy library once, not a second time, as it would be the case with one consisting of say,  FLAC and MP3.


it's actually very easy to manage tags across a lossless/lossy collection using foobar2000. you can copy/paste from one set of files to another with ease. obviously it's imperative that each set are in the exact same order. i'm pretty sure only it only updates files where differences have been found, even if you do you whole collection at once.

http://wiki.hydrogenaudio.org/index.php?ti..._sets_of_tracks

Hence my considering WavPack as the best of both worlds when it comes to lossless encoding (and just to stick to the now-ancient OP).


yep, but i suspect a large chunk of people won't have that choice because of the portable device they use.
  • Last Edit: 04 April, 2014, 09:49:25 AM by marc2003

Which is the best lossless codec?
Reply #302
obviously it's imperative that each set are in the exact same order.

IMO, that alone, deems this solution less practical than WP's.

i'm pretty sure only it only updates files where differences have been found, even if you do you whole collection at once.

That I didn't know. Thanks.
Listen to the music, not the media.

Best,
Nilson

  • eahm
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #303
Now that the table (http://wiki.hydrogenaudio.org/index.php?title=Lossless_comparison#Comparison_Table) is updated, why ID3/APEv2 is worst than APEv2 or Vorbis tags?

  • ktf
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #304
why ID3/APEv2 is worst than APEv2 or Vorbis tags?

Please read this post in this topic a few weeks back.

Having two allowed (by which I mean the official software and plugins support it) tagging schemes means that you can have two sets at once. Different versions of ID3 alone can give problems, see this topic, especially post #6. I think it's quite a downside to have two possibly conflicting versions of tags in one file.
  • Last Edit: 11 April, 2014, 05:08:26 AM by ktf
Music: sounds arranged such that they construct feelings.

  • ktf
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #305
For anyone wondering, as a reminder this thread accompanies this wiki article: http://wiki.hydrogenaud.io/index.php?title...less_comparison

I've noticed some editing which I do not find reasonable, so I'd like to discuss it. A while ago, I removed TTA from the main section of the mentioned wiki page because it doesn't seem to attract much attention (at least not at HA). The edit summary:  Stripped table from and shortened text about Shorten, LA, TTA, ALS, SLS and Real Lossless

Recently, a TTA dev (Ald) has added it to the table again, but added an extra row with the feature "Password protection". I subsequently removed this, because it is only a minor feature that is supported by only one codec. There are other features much more noteworthy like cuesheet embedding and having an MD5 hash for security. Furthermore, I sorted the table on popularity.

Apparently the TTA dev didn't agree, because the changes were undone. Furthermore, the flexibility of TTA was named 'adaptive', which is just an eufemism for not having any options, which is what the flexibility means.

What do you think? Should I go ahead and undo Ald's undo-edits? It starts to look like edit-warring
  • Last Edit: 10 November, 2014, 11:21:32 AM by ktf
Music: sounds arranged such that they construct feelings.

  • lvqcl
  • [*][*][*][*][*]
  • Developer
Which is the best lossless codec?
Reply #306
Furthermore, the flexibility of TTA was named 'adaptive'


Yes, this surprised me when I saw it several days ago.

  • tuffy
  • [*][*][*]
Which is the best lossless codec?
Reply #307
Apparently the TTA dev didn't agree, because the changes were undone. Furthermore, the flexibility of TTA was named 'adaptive', which is just an eufemism for not having any options, which is what the flexibility means.

I'm not sure I'd rate "flexibility" on a good or bad scale the same way that decoding speed or compression ratio are.  FLAC encoding has a lot of tunable parameters whereas ALAC has nearly no tunable parameters and TTA has none at all.  But I wouldn't consider the presence of lots of possible encoding knobs to necessarily be a virtue or the lack of them to be a fault.

  • ktf
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #308
But I wouldn't consider the presence of lots of possible encoding knobs to necessarily be a virtue or the lack of them to be a fault.

Why?

If you don't like buttons, you can use all codecs without them. FLAC will default to compression level 5. So I wonder, how do you think having more options, and thus more flexibility, can be a bad thing? Just because it might frighten people to do something wrong?

I agree, that a codec like OptimFROG has so many options and combinations of them (if I set mode, should I set optimize as well? Which combination is best? etc.) but most codecs have a relatively simple system for this.
Music: sounds arranged such that they construct feelings.

  • tuffy
  • [*][*][*]
Which is the best lossless codec?
Reply #309
Why?

If you don't like buttons, you can use all codecs without them. FLAC will default to compression level 5. So I wonder, how do you think having more options, and thus more flexibility, can be a bad thing? Just because it might frighten people to do something wrong?

I agree, that a codec like OptimFROG has so many options and combinations of them (if I set mode, should I set optimize as well? Which combination is best? etc.) but most codecs have a relatively simple system for this.

I just think that the effects of flexibility are more important than the presence of flexibility.  Like it's noteworthy that FLAC offers a tradeoff between encoding speed and compression ratio, but I wouldn't hold a lack of options against some hypothetical codec that doesn't offer any tune-able parameters but compresses very well.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
Which is the best lossless codec?
Reply #310
What if a codec is faster at both encoding and decoding and compresses better, but has no options?

Some codecs have so many options, to the point that it's far more confusing than effective, especially when some new codec comes along and blows it out of the water in terms of performance.

IMO, that row should either be reworked or removed. That it requires a note to explain what the title means is ridiculous.
  • Last Edit: 11 November, 2014, 11:46:51 AM by greynol
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • ktf
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #311
I just think that the effects of flexibility are more important than the presence of flexibility.

Those possible effects are already mentioned in the table: compression and speed are mentioned at the top. TTA chose to be inflexible, and I might compress better/faster because of this tradeoff (less development time needed tuning, less format header for different modes etc.) So, yes, not being flexible might be a advantage for a codec, but that advantage is already clearly visible someplace else in the table.

Quote
but I wouldn't hold a lack of options against some hypothetical codec that doesn't offer any tune-able parameters but compresses very well.

In that case, just ignore that row in the table. That is no reason not to leave it in for others to take into account, right?

edit:
What if a codec is faster at both encoding and decoding and compresses better, but has no options?

In that case, those benefits are already mentioned in the table.

Quote
IMO, that row should either be reworked or removed. That it requires a note to explain what the title means is ridiculous.

What about naming it presets and mentioning the number of presets that are available (for FLAC that would be 9, WavPack 4 or 4*6 if you count the x options, Monkey's 5, TTA 1 etc)
  • Last Edit: 11 November, 2014, 11:58:17 AM by ktf
Music: sounds arranged such that they construct feelings.

  • greynol
  • [*][*][*][*][*]
  • Global Moderator
Which is the best lossless codec?
Reply #312
With no color coding, yes, I think that would be an improvement.

I agree that the password row should be removed. I had similar reservations about 1-off features added to the secure ripper comparison page.

I often wonder:
What good are rows where there is no variation between participants?
13 February 2016: The world was blessed with the passing of a truly vile and wretched person.

Your eyes cannot hear.

  • Porcus
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #313
I often wonder:
What good are rows where there is no variation between participants?


Depends on whether everybody knows already, that they need not spend time digging up that piece of information.

  • rosbifmark
  • [*]
Which is the best lossless codec?
Reply #314
Following on from this discussion, in particular regarding error detection:

If I first rip a CD (with errors) to an ALAC file, I understand:

-there is no checksum error detection.
-with errors, the ALAC file may stop playing.

If I then convert the ALAC file to a FLAC file:

-will the new FLAC file have the checksum data present (or have I lost this capability since the file was once ALAC)?
-will the new FLAC file be able to play through the errors  (or again since it was once ALAC, will this capability be lost)?

This will help me decide if I can encode in ALAC and then at a later date, convert to FLAC (and re-gain error detection capabilities).

Cheers

Which is the best lossless codec?
Reply #315
If I first rip a CD (with errors) to an ALAC file, I understand:
You mean if the ALAC file is corrupted somehow, right? Your statements fit that scenario, so I'll assume that's what you are referring to.

If I then convert the ALAC file to a FLAC file:

-will the new FLAC file have the checksum data present (or have I lost this capability since the file was once ALAC)?
-will the new FLAC file be able to play through the errors  (or again since it was once ALAC, will this capability be lost)?
The new FLAC file will have a checksum, but the checksum will match the corrupt audio and not the original CD. You will be able to detect if any further corruption affects the FLAC file, but not corruption from before it was encoded to FLAC.

The new FLAC file won't have any errors to play through. It will contain a perfect copy of whatever audio you are able to get out of the ALAC file. If the ALAC file cuts short due to an error, the FLAC file will simply end there.

  • lvqcl
  • [*][*][*][*][*]
  • Developer
Which is the best lossless codec?
Reply #316
Furthermore, the flexibility of TTA was named 'adaptive'

Yes, this surprised me when I saw it several days ago.

Also very interesting TTA property is its "High linearity". 

  • Porcus
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #317
If I first rip a CD (with errors) to an ALAC file, I understand:

-there is no checksum error detection.
-with errors, the ALAC file may stop playing.


You mean, the CD has errors making the ripping process stop midway in a track? Then no matter what format you choose, you will get a technically valid signal which is whatever the ripper managed to get out of the CD (with all those errors).

  • ald
  • [*][*][*]
  • Developer
Which is the best lossless codec?
Reply #318
A while ago, I removed TTA from the main section of the mentioned wiki page because it doesn't seem to attract much attention (at least not at HA). The edit summary:  Stripped table from and shortened text about Shorten, LA, TTA, ALS, SLS and Real Lossless


Hi, KTF! Why are you trying to remove TTA project? If you don't like it? Why?
You wrote: "Sorted table on popularity", but it's not true. I have no such statistics.
The codec have a good download statistics at Sourceforge. TTA codec still popular in Russia and Japan. You can easily found TTA files on russian torrents, but there is no files in TAK or OptimFROG formats for example.
Where did you get information about the popularity of codecs?

Recently, a TTA dev (Ald) has added it to the table again, but added an extra row with the feature "Password protection". I subsequently removed this, because it is only a minor feature that is supported by only one codec.


Again, you have removed the "Password protection" feature from table without discussing here, why? Please be patient.
TTA codec has three significant features:

- Password protection;
- Highest encoding speed;
- Ultra low latency.

The current version of the table does not contain any of this.
You have explained it by this way: "it's only a minor feature". Please note, that this is just your opinion. Many people in the world think otherwise.
And yes, you are right, "Password protection" feature that is supported by only one codec.
This information should be in the Lossless comparison table if you want to compare codecs correctly.

Anyway, thanks for your work!

  • Porcus
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #319
Again, you have removed the "Password protection" feature from table without discussing here, why? Please be patient.


Until such a feature has been discussed and there is some kind of consensus that it is important enough to signify a separate line in such a table, I think one should refrain from adding it. And if someone just adds it without discussion, I think the appropriate thing would be to revert the change, then discuss, and then if agreed upon, enter it.

And especially I would say this applies when a developer of one of the formats throws in a feature-line that is unique to their own product, based on their own assessment of the significance of the features of their own products. You are not by any means neutral, hm? ;-)


  • lvqcl
  • [*][*][*][*][*]
  • Developer
Which is the best lossless codec?
Reply #320
- Highest encoding speed;

According to the lossless codecs comparison by ktf, TTA encoding speed is 150x realtime, FLAC -4 is 300x, TAK -p0 is 400x. So, high but not the highest.

- Ultra low latency.

How did you measure it? 

  • ktf
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #321
Hi, KTF! Why are you trying to remove TTA project? If you don't like it? Why?

Because there is next to no discussion about it here at HydrogenAudio, and because I haven't seen a TTA file it in the wild, ever.

Quote
Where did you get information about the popularity of codecs?

Here: http://www.hydrogenaud.io/forums/index.php?showtopic=105188

Quote
Again, you have removed the "Password protection" feature from table without discussing here, why?

Because it is a minor feature. Other (in my opinion more important) features, like MD5 checksumming, cuesheet support, having multiple encoding/decoding implementations, supporting 32-bit floats aren't listed either, because the table would become to large and therefore harder to read.

Quote
TTA codec has three significant features:

- Password protection;
- Highest encoding speed;
- Ultra low latency

As said, I don't consider password protection a major feature, TTA hasn't got the highest encoding speed and for that low latency, I'd like to see some numbers in context, as I don't think it is unique in that sense.

Quote
Many people in the world think otherwise.

No one reading this thread has agreed with you yet.

Quote
This information should be in the Lossless comparison table if you want to compare codecs correctly.

If we include all information on features, the table gets too big. A line has to be drawn somewhere.
  • Last Edit: 11 February, 2015, 12:18:27 PM by ktf
Music: sounds arranged such that they construct feelings.

  • Porcus
  • [*][*][*][*][*]
Which is the best lossless codec?
Reply #322
I think that password protection is a neat feature. Just because I think it is "neat" it doesn't mean it is significant enough.

Let me just for illustration point out that the zip format - PKWare's version, that is - has since 2007 supported WavPack as its Compression Method 97. There you go, password protection. If anyone cares.

And if anyone cares, .zip as audio format isn't that hard to implement. I do not know whether it is implemented in practice - those compression methods are not universally compatible, and I do not bother to check whether VLC (or fb2k for that matter) supports Compression Method 97.

And if anyone cares, WavPack-in-zip could even be mentioned in the text in the wiki. (WavPack in .zip container isn't that unlike whatever-in-Matroska, is it?)

Now is there anyone who thinks this is worth a line in an overview table?


Other (in my opinion more important) features, like MD5 checksumming, cuesheet support, having multiple encoding/decoding implementations, supporting 32-bit floats aren't listed either, because the table would become to large and therefore harder to read.


... piping ... unicode ...

  • ald
  • [*][*][*]
  • Developer
Which is the best lossless codec?
Reply #323
- Highest encoding speed;

According to the lossless codecs comparison by ktf, TTA encoding speed is 150x realtime, FLAC -4 is 300x, TAK -p0 is 400x. So, high but not the highest.


Please compare codecs test results with similar compression ratio:
http://www.squeezechart.com/audio.html

- Ultra low latency.

How did you measure it? 


Lower latency of the compressor is achieved by fully adaptive coding algorithm, without buffering stage, which is required in many codecs for the preliminary analysis of the encoded block of input data. The size of the buffer completely determines the delay. The TTA codec algorithm uses a single buffer of 32 bits for the formation of entire bytes from a bit sequence of variable length codes at the codec output. The buffer of this size means nearly zero delay in the encoding process. The latency of the codec was measured by my colleagues and it's value is about 0.1 ms. Note, that this value is not for console version of the codec, because the console version has a big buffers for read-write operations.

  • ald
  • [*][*][*]
  • Developer
Which is the best lossless codec?
Reply #324
Until such a feature has been discussed and there is some kind of consensus that it is important enough to signify a separate line in such a table, I think one should refrain from adding it.


Sorry, I did not know about this topic before.
No problem. I will wait for results of discussion :-)