Question About MP3s

[bostonman]

Over the years I've downloaded MP3s. In the past I'd listen to them and/or burn them to a CD. From what I remember, the MP3, when played on a PC would sound fine, but I'd hear a blip on the CD. Upon further investigating, I'd notice the MP3 contained a flaw and it wasn't as obvious when played as a file.

My guess at the time: subtle errors in the file would be emphasized when playing it on a CD

With the advent of USB ports in car radios, I've now been using a thumbdrive to play MP3s, however, I've noticed errors in a few MP3 files.

With this being said, I'm curious why these errors occur. I assume when the MP3s are being downloaded that CRC is being done so the files should match the source. If this is the case, would it be exclusively due to errors done by the creator?

Also, why are some MP3s louder than others? If the MP3 is being converted, should the volume always be the same since it's just digital (ignoring someone tampering with the file)?

[Siwastaja]

Who knows? It's obvious that something is broken in your chain, but having gazillion cheap consumer devices, all running software or firmware developed in a rush, in a chain produces gazillion of opportunities for things to go wrong.

99% likely, it's just a bug somewhere; impossible to say where. It can be, for example, but not limited to:

* Original digitization
* File corruption in the uncompressed original file
* Buggy MP3 encoder
* File corruption on the web server /
* File corruption in the P2P software (CRC doesn't help if the validation happens to be buggy!)
* File corruption on your hdd/ssd; physical or filesystem bug;
* Buggy USB stick (surprisingly common)
* Buggy CD burner software
* Physical damage on CD
* Buggy data CD reading mode in the car radio /
* Buggy USB implementation in the car radio
* Buggy MP3 decoder in the car radio

You can try elimination, for example try playing the same USB stick or CD on another player and see if it behaves similarly. Or are the glitches you see always at the same time? Such clues help you narrow down potential suspects.

What comes to loudness, volume is obviously decided by whoever created the mp3 file. mp3 stores digital audio data with (typically) 16 bits of resolution which leaves quite a lot of headroom to only use part of the range, or use it fully, or even "overuse" it applying compression or other similar techniques.

[bostonman]

These MP3 questions date back to my old computer, old car stereo, etc... and are carrying over into everything I have now.

When a song on a CD that was converted from a downloaded MP3 would blip, I'd check the original MP3 wave in a program and would see a blip in the wave. Upon listening closer on the computer (listening as a regular MP3) the subtle blip would be heard.

I attributed the louder blip on the CD to be the software not handling the blip in the MP3 and translating it into something worse.

From what I can tell, any MP3 errors originate from the original MP3 file. I've never had to re-record a CD because it had a blip and the MP3 didn't.

My main question is regarding whether an MP3 can be downloaded incorrectly from the source where the source is fine, but the CRC (or whatever error checking methods are done now) fails thus providing an altered MP3.

From experience, I'd say a good 10% of MP3s I've downloaded have blips.

[andy3055]

If you compare downloaded file to what you have on the CD, you will be able to make a determination as to where the blip is. You say they are old. How old? There was a time when burning CDs would be interfered by other apps running at the same time, introducing errors.
Just a thought.

[wraper]

CD with MP3 is a usual data disc which should not read with any errors unlike audio CD where error correction is not strong and some lost data is acceptable and is actually normal.

[SiliconWizard]

From what I can tell, any MP3 errors originate from the original MP3 file. I've never had to re-record a CD because it had a blip and the MP3 didn't.

My main question is regarding whether an MP3 can be downloaded incorrectly from the source where the source is fine, but the CRC (or whatever error checking methods are done now) fails thus providing an altered MP3.

From experience, I'd say a good 10% of MP3s I've downloaded have blips.

From what I've seen and what is most likely: this comes from errors either from the original audio file, or from the MP3 encoding itself.

If your MP3s are relatively old, chances are many, if not most of them were created this way:
- Extracted audio from some audio CD. This part, depending on the software, CD-ROM drive and drivers you used, could occasionally cause "dropped" samples. Sometimes it would also just come from the CD itself. Audio CDs can be a bit scratched and yield dropped audio samples. Audio CD players are designed to compensate so this is usually not audible when played on one, but extracted audio, usually doesn't. The drops will just appear as either zeros or some random values. CD-ROM drives + extracting software very rarely compensated for dropped samples. This has probably significantly improved over the years.
- The MP3 encoder could be either buggy, or it could also introduce various kinds of nasty noises every once in a while when it was not registered! Yes, MP3 encoders back in the days were often NOT free.

[CatalinaWOW]

From what I can tell, any MP3 errors originate from the original MP3 file. I've never had to re-record a CD because it had a blip and the MP3 didn't.

My main question is regarding whether an MP3 can be downloaded incorrectly from the source where the source is fine, but the CRC (or whatever error checking methods are done now) fails thus providing an altered MP3.

From experience, I'd say a good 10% of MP3s I've downloaded have blips.

From what I've seen and what is most likely: this comes from errors either from the original audio file, or from the MP3 encoding itself.

If your MP3s are relatively old, chances are many, if not most of them were created this way:
- Extracted audio from some audio CD. This part, depending on the software, CD-ROM drive and drivers you used, could occasionally cause "dropped" samples. Sometimes it would also just come from the CD itself. Audio CDs can be a bit scratched and yield dropped audio samples. Audio CD players are designed to compensate so this is usually not audible when played on one, but extracted audio, usually doesn't. The drops will just appear as either zeros or some random values. CD-ROM drives + extracting software very rarely compensated for dropped samples. This has probably significantly improved over the years.
- The MP3 encoder could be either buggy, or it could also introduce various kinds of nasty noises every once in a while when it was not registered! Yes, MP3 encoders back in the days were often NOT free.

I suspect this is the essence of the answer.  When I rip scratched CD's it is obvious that sometimes the ripping software is repeatedly rereading the same sectors of the CD attempting to get an error free read.  Sometimes it gives up and announces that the CD is damaged.  Sometimes it goes on after a variable number of attempts.  And frequently the resulting MP3 has audio flaws.   But not always.   I have learned to listen to the file if I notice those multiple reads.  In almost all cases your MP3's were ripped from a CD somewhere along the line, and whoever did it may not have been diligent in their methods.

If you still have access to the CD I have found the headlight polishing kits used to remove yellowing and scratches from plastic headlight lenses highly effective in correcting headlight scratches and eliminating those read errors.  It only takes a couple of minutes using the finest grit sandpaper (typically 2000 or 3000) and then following up with the polishing compound.

There are any number of freeware programs available that will go through a directory or directory tree of MP3s and normalize the volume.  Just watch out for malware payloads.

[bostonman]

In general, when a MP3 is downloaded, a CRC is used, correct?

When I stated I burned MP3s to a CD in a previous message, I should have said I was creating audio CDs (this was years ago). When I heard a blip in a song, I'd examine the MP3 wave in a music program and would always see an error in the wave indicating it was either downloaded with errors or the original person who ripped the CD caused the error.

I've basically always found the error is before my ownership; whether a CRC error took place or the original person caused an error.

If an MP3 (and other information) is always checked via CRC, and I can basically abruptly stop downloading an MP3 and continue another time, and that wouldn't affect the information being transferred, then I guess it's the original person who ripped the CD.

[CatalinaWOW]

In general, when a MP3 is downloaded, a CRC is used, correct?

When I stated I burned MP3s to a CD in a previous message, I should have said I was creating audio CDs (this was years ago). When I heard a blip in a song, I'd examine the MP3 wave in a music program and would always see an error in the wave indicating it was either downloaded with errors or the original person who ripped the CD caused the error.

I've basically always found the error is before my ownership; whether a CRC error took place or the original person caused an error.

If an MP3 (and other information) is always checked via CRC, and I can basically abruptly stop downloading an MP3 and continue another time, and that wouldn't affect the information being transferred, then I guess it's the original person who ripped the CD.

I think we said the same thing in different words.

[Siwastaja]

File corruption is always possible and can happen anywhere.

If the mp3 file is downloaded from a web server through TCP, which is what normally happens on "normal" web page downloads, TCP itself guarantees no data corruption in network transfer. TCP layer implementations are notably almost definitely bug-free, they are used and stress-tested so widely. This is the least likely place for corruption to happen.

Peer-to-peer software may use TCP or UDP, and definitely most of them implement some checksumming of transfers, but such programs are at higher risk to contain bugs.

AFAIK, MP3 file format itself does not contain CRC or error correction, but I might be wrong.

Filesystems are usually fairly robust, but who knows, anything can happen.

Buggy software is still #1 contributor to problems.

I have personally seen a Windows filesystem corrupt a bunch of mp3s so that one particular mp3 played a very interesting collection of clips from other songs. The result was artistically so interesting it didn't bother me. Or I don't know, maybe it wasn't the filesystem, maybe it was some piece of buggy userland software gone berserk, modifying random files.

[CatalinaWOW]

There are also rumors that the copyright protection crowd created worms that would corrupt files that didn't incorporate DRM features.  I saw some evidence of this actually happening quite a few years ago, but it probably wasn't actually the corporations involved because of the liability and publicity blowback that could occur.  It is possible that some of that happened and is still floating around out there.

[Nominal Animal]

In the early times, not all drives were able to read audio CDs without introducing errors at the higher ripping rates.  (Where I live, we pay copyright levies on empty media, and have the legal right to do that.)

There were dedicated software, but even with those, not all CD drives performed equally well.  Occasional "glitches", errors in the ripped data, were common.  Most people tended to ignore those – much like they ignored typical playback issues with VHS tapes, that nowadays would probably drive people used to perfect playback crazy –, so I'd guess the most likely issue with older mp3's happened at the CD ripping stage.

[tooki]

Over the years I've downloaded MP3s. In the past I'd listen to them and/or burn them to a CD. From what I remember, the MP3, when played on a PC would sound fine, but I'd hear a blip on the CD. Upon further investigating, I'd notice the MP3 contained a flaw and it wasn't as obvious when played as a file.

My guess at the time: subtle errors in the file would be emphasized when playing it on a CD

With the advent of USB ports in car radios, I've now been using a thumbdrive to play MP3s, however, I've noticed errors in a few MP3 files.

With this being said, I'm curious why these errors occur. I assume when the MP3s are being downloaded that CRC is being done so the files should match the source. If this is the case, would it be exclusively due to errors done by the creator?

Also, why are some MP3s louder than others? If the MP3 is being converted, should the volume always be the same since it's just digital (ignoring someone tampering with the file)?

There’s no reason a CD would “emphasize” errors. More likely you just had better speakers on the CD player than the computer.

By “download”, you mean from places other than official music stores? Well, you never know when and how a pirated MP3 was created — was the original CD undamaged and clean? Was the drive working well? Was error correction applied? But even assuming it was, a lot of pirate sharing systems involve swarm file sharing. Usually it works, but it can happen that pieces are stuck together wrong. And then those files get re-shared.

As for volume, what makes you think the original uncompressed source has the same volume as everything else? The average volume of music has been rising for years. Google “loudness wars”.