General Questions about CD-Ripping Programs

[Bespin1138]

Out of curiosity, I have some general questions about CD ripping programs. Over the years I have used various programs to rip my CDs. These include Windows Media Player, Creative Labs, iTunes, Music Match Jukebox, Real Media Player, Winamp, Media Monkey, and currently dBpoweramp. Now, through some help from the kind users on this forum, I have learned that when I rip a CD to a lossless format such as FLAC, the file will be identical regardless of whether or not the PC hardware components such as the optical drive, CPU, or hard drive are the same or different.

I feel that I should better understand the software side of CD ripping as well and felt it was better to ask this as a separate topic. Now, I am aware that certain programs, at least in the past, such as Windows Media Player and iTunes, did not have secure rip settings or accurate rip verification options but other programs like Media Monkey and dBpoweramp do. My questions are as followed:

1. I know that each CD ripping program is written differently. I am also aware that some CD ripping programs use different meta data tags and settings such as Artist, Album, and Genre. Now if I used a program that had both secure rip and accurate rip or other accuracy verification settings, and ripped my CDs to FLAC, would I get FLAC audio files that are identical or different in terms of audio quality and/or audio properties if I were to rip the same CD using Winamp, Media Monkey, or dbPoweramp?

2. If I were to use the same programs previously mentioned when I ripped the CDs to FLAC and made copies of the songs in lossy formats, such as MP3 or AAC, would I get audio files that are identical or different in terms of audio quality and/or audio properties?

3. If I used one program to rip my CDs to FLAC and years later I switch to a different program and use it to create lossy copies like MP3 or AAC, would the audio quality and/or audio properties be identical or different from the files that would have been created if I had stuck with the same program that I used to rip my CDs to FLAC?

I appreciate the information.

Thanks.

[garym]

Re: General Questions about CD-Ripping Programs

1. I know that each CD ripping program is written differently. I am also aware that some CD ripping programs use different meta data tags and settings such as Artist, Album, and Genre. Now if I used a program that had both secure rip and accurate rip or other accuracy verification settings, and ripped my CDs to FLAC, would I get FLAC audio files that are identical or different in terms of audio quality and/or audio properties if I were to rip the same CD using Winamp, Media Monkey, or dbPoweramp?

They would all be identical in terms of the audio content/sound quality, assuming there were no errors in the ripping. For example, I'm not sure what sort of error identification process Winamp had. Or Media Monkey for that matter. So absent ripping errors, they would all have identical audio content. Each program may write different tags, some might use tag fields that others don't use. But this would only change the metadata tag info. It would have zero impact on the audio content of these lossless FLAC files.

2. If I were to use the same programs previously mentioned when I ripped the CDs to FLAC and made copies of the songs in lossy formats, such as MP3 or AAC, would I get audio files that are identical or different in terms of audio quality and/or audio properties?

Here there could be differences in the audio content. The codecs that create mp3 or aac files are all a bit different and handle things a bit different. With a quality program it is unlikely that these various files would SOUND different, but the audio content is unlikely to be bit perfect. Even with mp3, there are different endcoders (LAME, fraunhoffer, others....). Same with AAC and m4a, etc. So once you go to lossy files, nothing is identical (but also not likely a problem either).

3. If I used one program to rip my CDs to FLAC and years later I switch to a different program and use it to create lossy copies like MP3 or AAC, would the audio quality and/or audio properties be identical or different from the files that would have been created if I had stuck with the same program that I used to rip my CDs to FLAC?

Really the same answer as *2 above. The audio content of FLAC = FLAC = FLAC no matter what program was used to create the FLAC file. Now once you convert to lossy files (mp3, etc.) then it depends on the particular codec, the version of the codec, etc. But if you're asking "do I have to worry that I am converting FLAC files to mp3 with dbpa, even though I created the FLAC file with Media Monkey, the answer is NO. Nothing to worry about. EXCEPT, you can't be 100% sure that the original FLAC file was ripped without error if it didn't have an ACCURATERIP match when ripped, or wasn't ripped with some sort of "secure" ripping process (e.g., originally iTunes had no secure ripping process). But even with this caveat, keep in mind that *most* CDs ripped with *most* modern rippers (last 15 years or so), even if not secure rippers, still produce the same audio content in the digital file. That's because most CDs don't produce errors when ripping. But the key is, for the few that do, we'd like to know!

And by the way, the partner program to dbpa, PerfectTunes, is perfect for checking whether your OLD rips from other programs match the AccurateRip database. One can use PerfectTune to batch check your entire lossless collection to see whether your digital files match the accuraterip database.

I hope that helps.

[Bespin1138]

Re: General Questions about CD-Ripping Programs

Thanks for the answers. I will definitely keep that PerfectTunes program in mind.

[schmidj]

Re: General Questions about CD-Ripping Programs

To answer your questions:

FLAC and other "lossless" codecs store a file which (if the CD is ripped without error) can be played as a bit accurate copy of the original data on the CD. You can also convert a FLAC file to a WAV file (which is a direct representation of the bits on the CD track) or any other "lossless" codec or visa versa "losslessly", the bits played back are identical.

FLAC and other lossless codec storage formats are like "zip" files of data, but are optimized to take less storage space with the typical bitstrings that exist in music files. You can "zip" a music file just like a word document, and unzip it and play it bit perfect, it is just that zip was optimized for other types of files and a zipped music file will probably be larger than a file created by FLAC or other lossless music codec.

Now the big "if"... If the file is ripped without error... (And if the file, after ripping doesn't get corrupted.) then any properly working ripping program will give you an identical FLAC file, and any properly working player will play that file back identically, as we say, bit accurate. But errors do happen. The CD has a scratch or even a fingerprint. It is warped or otherwise damaged. It could even be a defective pressing from the factory. Even bumping the CD drive while ripping may create errors. More common than you might think. Many bit errors are inaudible. The ear is just not that sensitive, even the "golden ears". Other bit error may be audible but not really annoying, a click like you used to hear thousands of when playing a record. But other errors may make the rip "unlistenable", it skips, drops out, ends abruptly in the middle of the track, or has more pops and clicks than you can bear to listen to. And with most rippers, you won't know until you play back the ripped track and listen to it.

But dBpa and a few other rippers use a technique to attempt to verify that the rip is truly a lossless representation of the CD track. Some, including dBpa use a mathematical technique called the checksum. (The same technique is used throughout the digital world to verify that data doesn't become corrupted, every TCP packet includes a checksum and the receiver calculates the checksum from the packet data and makes sure the calculated checksum matches the transmitted checksum. If it doesn't the packet is sent again.)

The checksum is calculated for each track ripped by dBpa. Unfortunately, the designers of the CD did not include a checksum in the audio data on the manufactured CD. But dBpa does the next best thing. When you rip CDs using dBpa, the software saves the checksums it calculated for each CD together with data it uses to identify the CD. Every now and then (usually monthly), dBpa calls "home" and transmits all the checksums to a database at the dBpa headquarters. Then when someone else rips another copy of the same CD, the dBpa software queries the database, looks to see if it has checksums for that particular CD, and checks whether the checksum matches. If it does, the rips match, with an error rate of less than one in a million. (Spoon would have to tell you the actual error rate for the checksum he calculates, but it is infinitesimally small.) So if the checksums (which are displayed on your computer screen) match, you have a bit accurate rip, and if you ripped to a lossless codec, it will play back absolutely identically.

Now what if your CD isn't in the database? Uncommon for pop CDs, but fairly common for the ethnic CDs that I have many of. (No one else owning dBpa has ripped that same CD before me.) Well, then if you have set it up to do secure rips, dBpa rips the tracks multiple times. (You can select how many times in the ripping options depending on how paranoid you are.) If all the rips of a track calculate the same checksum, the chances are pretty good that you have an accurate rip, as damaged CDs typically play back differently each time they are played, and the checksums won't match. So your rip is marked as "secure", and the checksum data will be sent back to the central database the next time the software calls home.

If the checksums don't match from your multiple rips of a track, dBpa can be set to go one step further. It can keep track of which frames (1/75 of a second each) on the track don't match and try ripping that frame over and over again until it gets matching results for just that frame, save that moment of audio and go on to the next bad frame, if any.

I've found that reripped frames are often not bit accurate, even if they get a match to each other in the rerip process, but almost always the discrepancy is inaudible.

This whole process is what dBpa calls "accuraterip".

A few other rippers use a similar technique, either using dBpa's database or their own. Most do not, so you are much more likely to have defective rips in your collection with them.

Question two:

As we have said, any error free rip using any lossless codec will play back identically. But lossy codecs throw away some of the audio, to save storage space or transmission bandwidth. And once you throw it away, it's gone, you can't get it back. Most lossy codecs have settings with which you can determine how much audio data is thrown out. Typically, the more audio data you throw out the more likely you will hear a difference from the original. Some newer lossy codecs (or revisions of older codecs) are more "efficient" than some of the older ones. You can throw more out making for a smaller file or less bandwidth before most people will complain about the loss of quality with most source material. And your listening environment affects what you hear greatly also. You are much less likely to hear the "damage" to a compressed track listening to it in a noisy automobile interior than when listening to a good quality reproduction system in your living room or bedroom.

You'll find plenty of suggestions on how small you can make your files (or bitstream) using a given codec before it is audible. But each person will get a different number, and each person is willing to accept different amounts of audible degradation in their listening. Some people find listening to music on Alexia in their living room perfectly acceptable. Others find the need to spend thousands on high power amplifiers and fancy speakers. To each his (or her) own. And, for many years in the early 20th century, people were delighted to listen to scratchy 78RPM records with poor frequency response. It was that or nothing.

Personally, my collection exceeds 100,000 tracks. Most are FLAC, some are compressed formats, because they were downloads only available compressed, and some are from my first ripping, done to m4a for my Ipod and as of yet, not reripped (soon, I promise...) They live on a fairly large NAS server, backed up to a couple of large "brick" USB drives. A m4a copy lives on my phone, on a 500GB micro SD card, soon to become a terabyte when I can afford it. The FLAC files wouldn't come close to fitting on even a terabyte SD card. And actually, about 90 percent of my listening is in my car, from the phone via Bluetooth to my car radio. Historically, I could sometimes hear the difference between the lossless FLAC files and the more or less 120kB m4a files when played back to back on my home system, but I lost some of my hearing with illness a couple of years ago (I'm in my late 70s) so I doubt I could hear the difference now. But if you want to experiment, convert a few songs (best serious music, not pop) to various compressed formats at different bitrates and listen to the results compared to the lossless. See what your tolerance is.

Three: Probably different. First. it is unlikely that you'd remember exactly the settings you used to create the first copies. Second, new codecs and improvements to old ones happen from time to time. Third, there are different versions of some lossy codecs which will produce somewhat different results. Even dBpa gets updated copies of codecs from time to time. Some are bug fixes, some are operational improvements, like handling metadata better, but also some are improvements to the compression algorithms.

You should always keep (multiple copies of) your FLAC masters. Then you can make a copy converted to your lossy codec of choice from time to time, either using the dBpa converter or whoevers.

I used to use the dBpa ripper in multi-encoder mode to create FLAC and m4a at the same time. But I changed to ripping to just FLAC and then converting the entire library at the end of the day or week to create the m4a copy. Why? Because I'm always finding errors or typos in the metadata after I've ripped or fairly often in something I ripped years ago. If I use the multi-encoder, I have to fix those errors twice, once in FLAC, once in m4a. Easier to just fix the FLAC copies, then re-convert everything to fix the m4a copies. Yes, it takes quite a while to reconvert all the 100,000+ tracks, but I let it run all night.

Good luck, and spend as much time thinking about metadata as you are about the audio. That is where you'll find the most issues.

John