Hello,
i was wondering if there are some free tools to do batch inaudible watermarking on MP3 files? I am looking into possibilities of selling digital editions of music that will be released on my record-label, and i want to be able to watermark all the MP3's i am selling, to be able to trace them to the buyer if they show up on P2P networks etc if necessary.... please don't discuss the pro's and con's of watermarking and file-sharing...i just want to know if there are some free or cheap tools available that can do this kind of inaudible watermarking, and do this by batch..ie generate a new unique code and insert this watermark into seperate mp3s..
cheers, Olaf
Full Version: free batch watermarking tools for MP3?
you can "draw" frequencies with wavlab :-)
The short answer to your question is 'no.' The MP3 specification has no provision for watermarking. The RIAA might feel more kindly towards MP3 if it did. You might be able to place something in an ID3 tag, but it likely would not be worth the bother. You could distort the sound in some way, or include a few invalid frames. Whether or not this is inaudible I will let you decide. If you need some sort of copy protection or digital rights management Microsoft's Windows Media could be more appropriate.
QUOTE (woody_woodward @ Feb 17 2003 - 05:09 PM)
The short answer to your question is 'no.' The MP3 specification has no provision for watermarking. The RIAA might feel more kindly towards MP3 if it did. You might be able to place something in an ID3 tag, but it likely would not be worth the bother. You could distort the sound in some way, or include a few invalid frames. Whether or not this is inaudible I will let you decide. If you need some sort of copy protection or digital rights management Microsoft's Windows Media could be more appropriate.
ok clear answer...is there a free/cheap tool that includes inaudible watermarking into audio that survives the mp3 encoding? i know there are some very expensive ones that do just that..but i don't have the money to purchase those tools because it's all small-scale operations..
cheers, Olaf
A couple of possible suggestions:
1) You could first take your "master" MP3s (with no tags/additions whatsoever) and create an MD5 checksum for each. That way, even if someone added a tag, as long as the internal MP3 structure was not corrupted you could simply strip the tags from suspicious files and compare them to your checksums. Of course, you would also have to allow for the possibility of someone using MP3Gain to alter the relative volume of your MP3s, in which case you would need to know the original volumes of your files and also check the volume of suspicious files, adjusting accordingly.
2) If you are using L.A.M.E., you might be able to insert a spurious entry into the L.A.M.E./Xing tag. The advantage to this method is that the L.A.M.E. tag is going to survive MP3Gain'ing, but you would need to use a file-parsing program to search for your edited tags.
But - and I mean this with the best of intentions - I doubt anyone here would ever consider buying an album they knew was from an MP3 source. If it is your own record label, and recordings to which you own the copyright, you can of course do as you wish... but think about your customers. Is this truly what they would want, or be willing to purchase?
- M.
1) You could first take your "master" MP3s (with no tags/additions whatsoever) and create an MD5 checksum for each. That way, even if someone added a tag, as long as the internal MP3 structure was not corrupted you could simply strip the tags from suspicious files and compare them to your checksums. Of course, you would also have to allow for the possibility of someone using MP3Gain to alter the relative volume of your MP3s, in which case you would need to know the original volumes of your files and also check the volume of suspicious files, adjusting accordingly.
2) If you are using L.A.M.E., you might be able to insert a spurious entry into the L.A.M.E./Xing tag. The advantage to this method is that the L.A.M.E. tag is going to survive MP3Gain'ing, but you would need to use a file-parsing program to search for your edited tags.
But - and I mean this with the best of intentions - I doubt anyone here would ever consider buying an album they knew was from an MP3 source. If it is your own record label, and recordings to which you own the copyright, you can of course do as you wish... but think about your customers. Is this truly what they would want, or be willing to purchase?
- M.
QUOTE (olafmol @ Feb 17 2003 - 01:13 PM)
ok clear answer...is there a free/cheap tool that includes inaudible watermarking into audio that survives the mp3 encoding? i know there are some very expensive ones that do just that..but i don't have the money to purchase those tools because it's all small-scale operations..
No.
Because it will be defeated. See: http://cryptome.org/sdmi-attack.htm
You can't prevent a copy. If you mark it, the people will learn to remove the marks. Your attempts will only make your own costs go higher, all in vain. Time to redesign your business strategy. Nothing personal.
There is no point in working against your consumers needs, open the doors, see the reality. Time has changed. Make sure your clients get the message too, they will anyway, either the easy or the hard way. You choose.
Read this too: http://www.janisian.com/article-internet_debacle.html
Even artists know, so should you.
Will you treat your own customers as friends or enemies? Choose.
Will you lie to your client and sell them a system you know is going to be defeated?.
QUOTE (Artemis3 @ Feb 17 2003 - 09:05 PM)
Read this too: http://www.janisian.com/article-internet_debacle.html
Even artists know, so should you.
Even artists know, so should you.
Nice article.
It's nice to see Janis' honesty, but she is actually breaking the law herself:
"Additionally, when I buy a new CD, I make a copy for my car, a copy for upstairs, and a copy for my partner. That's three blank discs per CD."
That's actually illegal (in the USA anyway) according to most licenses that let you make a single backup copy for personal use.
Not that I care personally (or agree with the RIAA's stupid licensing terms), but it was interesting to note...
Watermarking is likely to NOT survive to lossy compression.
But you might want to have a look at something called Mp3Stego.
But you might want to have a look at something called Mp3Stego.
Watermarking mp3 files is possible, and it's free. See: http://www.cl.cam.ac.uk/~fapp2/steganography/mp3stego/
However I can see various problems about this approach:
- the mp3 encoder within mp3stego is very basic, thus you'll not get great sound quality,
- the hidden data will be destroyed if you transcode the mp3 file.
- the fact that mp3stego was used for encoding, should be fairly obvious when analyzing
The advantages of having a steganography-aware encoder:
- the original WAV data is untouched
- the encoder can take psychooacoustics into account to seamlessly store the hidden data.
I think the hidden data is stored as very small "details" in the MDCT coefficients of several (all?) frames of the mp3 file. If it's well done it could survive a mp3trim, but not a transcoding.
So in short, if you're in desperate need for steganography you can use this.. but if you're an audiophile, I suggest you use LAME and one of the following tricks:
- Make every mp3 encoding "unique", which means, encode each mp3 from a source file which has some bits swapped (64 changed bits in total should be enough). Keep track of the checksums of a few mp3 frames (or maybe the checksums of the bare mp3 data without tags and headers). Then distribute the files.
If people transcode (ie: re-encode) your files, you're screwed anyway. But otherwise, except for the obvious hate that file-sharing people will have for you, you can try it
Since many people re-tag their mp3's (and windows media player even does this automatically sometimes), this will require probably require some development for the integrity checking.
However I can see various problems about this approach:
- the mp3 encoder within mp3stego is very basic, thus you'll not get great sound quality,
- the hidden data will be destroyed if you transcode the mp3 file.
- the fact that mp3stego was used for encoding, should be fairly obvious when analyzing
The advantages of having a steganography-aware encoder:
- the original WAV data is untouched
- the encoder can take psychooacoustics into account to seamlessly store the hidden data.
I think the hidden data is stored as very small "details" in the MDCT coefficients of several (all?) frames of the mp3 file. If it's well done it could survive a mp3trim, but not a transcoding.
So in short, if you're in desperate need for steganography you can use this.. but if you're an audiophile, I suggest you use LAME and one of the following tricks:
- Make every mp3 encoding "unique", which means, encode each mp3 from a source file which has some bits swapped (64 changed bits in total should be enough). Keep track of the checksums of a few mp3 frames (or maybe the checksums of the bare mp3 data without tags and headers). Then distribute the files.
If people transcode (ie: re-encode) your files, you're screwed anyway. But otherwise, except for the obvious hate that file-sharing people will have for you, you can try it
Since many people re-tag their mp3's (and windows media player even does this automatically sometimes), this will require probably require some development for the integrity checking.
QUOTE (M @ Feb 18 2003 - 02:49 AM)
A couple of possible suggestions:
1) You could first take your "master" MP3s (with no tags/additions whatsoever) and create an MD5 checksum for each. That way, even if someone added a tag, as long as the internal MP3 structure was not corrupted you could simply strip the tags from suspicious files and compare them to your checksums. Of course, you would also have to allow for the possibility of someone using MP3Gain to alter the relative volume of your MP3s, in which case you would need to know the original volumes of your files and also check the volume of suspicious files, adjusting accordingly.
2) If you are using L.A.M.E., you might be able to insert a spurious entry into the L.A.M.E./Xing tag. The advantage to this method is that the L.A.M.E. tag is going to survive MP3Gain'ing, but you would need to use a file-parsing program to search for your edited tags.
But - and I mean this with the best of intentions - I doubt anyone here would ever consider buying an album they knew was from an MP3 source. If it is your own record label, and recordings to which you own the copyright, you can of course do as you wish... but think about your customers. Is this truly what they would want, or be willing to purchase?
- M.
1) You could first take your "master" MP3s (with no tags/additions whatsoever) and create an MD5 checksum for each. That way, even if someone added a tag, as long as the internal MP3 structure was not corrupted you could simply strip the tags from suspicious files and compare them to your checksums. Of course, you would also have to allow for the possibility of someone using MP3Gain to alter the relative volume of your MP3s, in which case you would need to know the original volumes of your files and also check the volume of suspicious files, adjusting accordingly.
2) If you are using L.A.M.E., you might be able to insert a spurious entry into the L.A.M.E./Xing tag. The advantage to this method is that the L.A.M.E. tag is going to survive MP3Gain'ing, but you would need to use a file-parsing program to search for your edited tags.
But - and I mean this with the best of intentions - I doubt anyone here would ever consider buying an album they knew was from an MP3 source. If it is your own record label, and recordings to which you own the copyright, you can of course do as you wish... but think about your customers. Is this truly what they would want, or be willing to purchase?
- M.
1) mmm.. a checksum is not exactly what i need....there is no need to check for tampering with the mp3..i only need to be able to trace "my" mp3's to the buyer, if necessary...i will look into the MD5 though...
2) isnt this tag editable for the user itself? or are these very different tags from IDv1 and IDv2?
to answer your last question, altough it might take away the focus somewhat from this discussion it's still interesting and something i thought about for very long, of course
i am not talking about albums here, firstly about seperate tracks..i just want to offer people that have no need for vinyl (like Finalscratch, CD and DJ software users) and it's additional costs of cutting, pressing, covers, printing, distribution, record-store et.etc. an alternative...the MP3s available for purchase are especially (re)mastered for the MP3 encoding (if necessary), they come from the original master-version and they will be encoded high-quality (256kbps or more)... prices will be a lot cheaper than the vinyl versions, i think around 3 euro for a track...and of course people who wish to can still buy the vinyl versions, either from the website or from your local recordstore (we have worldwide distribution).... as a DJ who switched from vinyl to CD's some years ago i always want to support the producers and labels (as i produce myself too) but i find it often a pity that i can only buy the (expensive, around 11 euro or more for import) vinyl version while i have no need for it's added values.....so i thought: put your money where your mouth is, practice what you preach etc.etc.... of course i am aware that piracy will always be there, in one way when one's music is spread on dedicated specialised music-sharing networks it's often a good way of promotion and a sign of the quality of the track...... but i also want to offer an alternative for people that want to pay for their music but don't want/need the added costs of a physical vinyl version.... i hope that makes things somewhat more clear
Olaf
QUOTE (Artemis3 @ Feb 18 2003 - 04:05 AM)
QUOTE (olafmol @ Feb 17 2003 - 01:13 PM)
ok clear answer...is there a free/cheap tool that includes inaudible watermarking into audio that survives the mp3 encoding? i know there are some very expensive ones that do just that..but i don't have the money to purchase those tools because it's all small-scale operations..
No.
Because it will be defeated. See: http://cryptome.org/sdmi-attack.htm
You can't prevent a copy. If you mark it, the people will learn to remove the marks. Your attempts will only make your own costs go higher, all in vain. Time to redesign your business strategy. Nothing personal.
There is no point in working against your consumers needs, open the doors, see the reality. Time has changed. Make sure your clients get the message too, they will anyway, either the easy or the hard way. You choose.
Read this too: http://www.janisian.com/article-internet_debacle.html
Even artists know, so should you.
Will you treat your own customers as friends or enemies? Choose.
Will you lie to your client and sell them a system you know is going to be defeated?.
as you can read my answer above this one, i do not want to fight the customer..don't confuse me with the evil RIAA or something like that..... i want to offer my buyers ways to choose....... i think it's time to rethink the business-strategy.. i think we agree...... but i just want to make sure if one of my customers might accidentally share the music he/she bought from me on a filesharing network, i can contact the buyer and let him/her know that the music that is payed for is distributed in illegal ways....... i am not trying to be some kind of spanish prosecution or something
Olaf
Another way how you could do it - it's about slight amplification:
I tried this: I took a test sample, apply short multiple short fadein-fadeout compinations with CoolEditPro (no dithering!), e.g. fadout 00:06.000/0dB -> 00:06.500/-0.01dB and fadein 00:06.500/-0.01dB -> 00:07.000/0dB.
Doing this at several positions of the sample I got a file that sounded identical to me even when trying hard to focus on the known positions. Now encoded both wit lame --aps, afterwards decoded, wav substraction... The result was digital silence, only at the positions where the fadeout/-in was applied there was some noise (up to 2 seconds, afterwards digital silence again). It's like a barcode.
If some1 transcodes/mp3gains a file treated like this the only chance to detect the watermark is to do the same to the original (-> use ecspot to guess transcoder...) but of course it's unsecure.
Second idea: After encoding remove some silence at the beginning/end of the track e.g. with mp3directcut and add a unique test signal which is the watermark instead which had been encoded at high level and is mp3gained to inaudible level afterwards (if it consists of very low + very high frequencies -70 - -80dB would be allright). I haven't tried this but I'm sure that when transcoding to mp3 at least some of the information will stay.
I tried this: I took a test sample, apply short multiple short fadein-fadeout compinations with CoolEditPro (no dithering!), e.g. fadout 00:06.000/0dB -> 00:06.500/-0.01dB and fadein 00:06.500/-0.01dB -> 00:07.000/0dB.
Doing this at several positions of the sample I got a file that sounded identical to me even when trying hard to focus on the known positions. Now encoded both wit lame --aps, afterwards decoded, wav substraction... The result was digital silence, only at the positions where the fadeout/-in was applied there was some noise (up to 2 seconds, afterwards digital silence again). It's like a barcode.
If some1 transcodes/mp3gains a file treated like this the only chance to detect the watermark is to do the same to the original (-> use ecspot to guess transcoder...) but of course it's unsecure.
Second idea: After encoding remove some silence at the beginning/end of the track e.g. with mp3directcut and add a unique test signal which is the watermark instead which had been encoded at high level and is mp3gained to inaudible level afterwards (if it consists of very low + very high frequencies -70 - -80dB would be allright). I haven't tried this but I'm sure that when transcoding to mp3 at least some of the information will stay.
QUOTE (NumLOCK @ Feb 18 2003 - 08:22 AM)
Watermarking mp3 files is possible, and it's free. See: http://www.cl.cam.ac.uk/~fapp2/steganography/mp3stego/
However I can see various problems about this approach:
- the mp3 encoder within mp3stego is very basic, thus you'll not get great sound quality,
- the hidden data will be destroyed if you transcode the mp3 file.
- the fact that mp3stego was used for encoding, should be fairly obvious when analyzing
The advantages of having a steganography-aware encoder:
- the original WAV data is untouched
- the encoder can take psychooacoustics into account to seamlessly store the hidden data.
I think the hidden data is stored as very small "details" in the MDCT coefficients of several (all?) frames of the mp3 file. If it's well done it could survive a mp3trim, but not a transcoding.
So in short, if you're in desperate need for steganography you can use this.. but if you're an audiophile, I suggest you use LAME and one of the following tricks:
- Make every mp3 encoding "unique", which means, encode each mp3 from a source file which has some bits swapped (64 changed bits in total should be enough). Keep track of the checksums of a few mp3 frames (or maybe the checksums of the bare mp3 data without tags and headers). Then distribute the files.
If people transcode (ie: re-encode) your files, you're screwed anyway. But otherwise, except for the obvious hate that file-sharing people will have for you, you can try it
Since many people re-tag their mp3's (and windows media player even does this automatically sometimes), this will require probably require some development for the integrity checking.
However I can see various problems about this approach:
- the mp3 encoder within mp3stego is very basic, thus you'll not get great sound quality,
- the hidden data will be destroyed if you transcode the mp3 file.
- the fact that mp3stego was used for encoding, should be fairly obvious when analyzing
The advantages of having a steganography-aware encoder:
- the original WAV data is untouched
- the encoder can take psychooacoustics into account to seamlessly store the hidden data.
I think the hidden data is stored as very small "details" in the MDCT coefficients of several (all?) frames of the mp3 file. If it's well done it could survive a mp3trim, but not a transcoding.
So in short, if you're in desperate need for steganography you can use this.. but if you're an audiophile, I suggest you use LAME and one of the following tricks:
- Make every mp3 encoding "unique", which means, encode each mp3 from a source file which has some bits swapped (64 changed bits in total should be enough). Keep track of the checksums of a few mp3 frames (or maybe the checksums of the bare mp3 data without tags and headers). Then distribute the files.
If people transcode (ie: re-encode) your files, you're screwed anyway. But otherwise, except for the obvious hate that file-sharing people will have for you, you can try it
Since many people re-tag their mp3's (and windows media player even does this automatically sometimes), this will require probably require some development for the integrity checking.
thanx for the link and info...looks interesting...but as you already suspected: audio quality is much more important than the watermarking to me..... as for the hate of the audio-sharing community: i don't care... non paying customers that aren't on my promo-list will of course always exist in these days, but i couldn't care less about them
i feel this is a littlebit like the open-source community i've been following for several years (i am a software guy)... people always talking about that they want open-source software because it let's them look into and modify the code, but in the end it's for 90% or more of the users the fact that the software is for free.... when they have to pay even a small amount they run away, or even worse: start cursing the software-creator as the next microsoft
anyway, i think i've made my point clear in my replies above..... you can agree, you cannot agree..it's all good.... if i never try we'll never see if it works....
Olaf
You're welcome.
Ok then.. you can use LAME --alt-preset standard for very good quality, and either:
- apply a slightly different scaling for each output mp3, using the "scale" switch in the LAME commandline,
- modify the LSB (least significant bit) of the ~ 128 first WAV samples, which will produce identical sounding, yet different mp3 files. Then, figure out some kind of integrity check to identify which mp3 was leaked on the net.
Cheers !
Ok then.. you can use LAME --alt-preset standard for very good quality, and either:
- apply a slightly different scaling for each output mp3, using the "scale" switch in the LAME commandline,
- modify the LSB (least significant bit) of the ~ 128 first WAV samples, which will produce identical sounding, yet different mp3 files. Then, figure out some kind of integrity check to identify which mp3 was leaked on the net.
Cheers !
Let's not forget that the intent is to watermark each individual mp3 in a unique way before it is sent to the buyer. It has to be a fast process which can be entirely automatic. I suspect this would take place on the web server or an application server.
What I suggest is:
1. store the raw (no tags) unaltered mp3 on a file server or database, and keep track of the maximum volume in a database. Tag information can be stored in that database as well. Number of frames might be useful as well for step 3.
2. prior to sending a mp3 to a client, run a batch process that:
3. randomly modify a copy of the mp3 very subtly: single step volume adjustment or reencode a single frame.
4. calculate the md5 for the resulting mp3 and store in the database
5. add the id3 tag
When you check a mp3 for watermarking, all you need to do is:
1. strip the mp3 of tags
2. check the max volume and use replaygain to adjust it back to the value store in the database is necessary
3. calculate the md5
4. check if the md5 matches anything in the db
This way your watermarking should survive tag manipulation and volume adjustment. Additionally you could also remove digital silence at the start and end of the original mp3 to make sure it also survives trimming.
I am not sure of which tools could help you automate the process but you might want to look at: mp3directcut, tag, mp3trim, mp3gain, md5sum/fsum.
IMHO adding stego or inaudible frequencies (see http://www.hydrogenaudio.org/forums/index....t=ST&f=1&t=1573 ) is not an audiophile acceptable solution; plus it does not lend itself well to automation.
What I suggest is:
1. store the raw (no tags) unaltered mp3 on a file server or database, and keep track of the maximum volume in a database. Tag information can be stored in that database as well. Number of frames might be useful as well for step 3.
2. prior to sending a mp3 to a client, run a batch process that:
3. randomly modify a copy of the mp3 very subtly: single step volume adjustment or reencode a single frame.
4. calculate the md5 for the resulting mp3 and store in the database
5. add the id3 tag
When you check a mp3 for watermarking, all you need to do is:
1. strip the mp3 of tags
2. check the max volume and use replaygain to adjust it back to the value store in the database is necessary
3. calculate the md5
4. check if the md5 matches anything in the db
This way your watermarking should survive tag manipulation and volume adjustment. Additionally you could also remove digital silence at the start and end of the original mp3 to make sure it also survives trimming.
I am not sure of which tools could help you automate the process but you might want to look at: mp3directcut, tag, mp3trim, mp3gain, md5sum/fsum.
IMHO adding stego or inaudible frequencies (see http://www.hydrogenaudio.org/forums/index....t=ST&f=1&t=1573 ) is not an audiophile acceptable solution; plus it does not lend itself well to automation.
QUOTE (arman68 @ Feb 18 2003 - 11:34 AM)
IMHO adding stego or inaudible frequencies (see http://www.hydrogenaudio.org/forums/index....t=ST&f=1&t=1573 ) is not an audiophile acceptable solution
It depends on how it is done !
Don't forget that all lossy audio codecs rely on increasing quantization noise where it isn't heard; for this reason you won't lose quality when you include encrypted data instead of regenerating white noise on playback.
Remember, one of the goals of encryption is to produce a data stream which is undistinguishable from a truly random source. This would fit the purpose greatly.
thanx for your replies! very helpfull!
arman68 is right about the automatic processing of the MP3.... of course for starters i can work out a system which i will feed with f.e. 20 MP3's i manually watermark and use these...but in the end it should be an automated and fast process.... i think the whole "hacking the watermark" issue is not very important though...because when people want they can probably find the MP3 easy on the file sharing networks or by a vinylversion and encode this to mp3.... the whole watermark thing should be seen as a control-thing....
i will check out all the info! thanx!
Olaf
arman68 is right about the automatic processing of the MP3.... of course for starters i can work out a system which i will feed with f.e. 20 MP3's i manually watermark and use these...but in the end it should be an automated and fast process.... i think the whole "hacking the watermark" issue is not very important though...because when people want they can probably find the MP3 easy on the file sharing networks or by a vinylversion and encode this to mp3.... the whole watermark thing should be seen as a control-thing....
i will check out all the info! thanx!
Olaf
Olafmol, if you want to automate things under windows, may I suggest you try batch files, using a command interpreter such as JPsoftware's 4DOS/4NT ?
QUOTE (arman68 @ Feb 18 2003 - 10:34 AM)
3. randomly modify a copy of the mp3 very subtly: single step volume adjustment or reencode a single frame.
does anyone have ideas/suggestions about how to approach this? i'm afraid i am not technical enough to directly understand how to inaudible change an mp3 frame..volume adjustments aren't really a good idea i think?
QUOTE (NumLOCK @ Feb 18 2003 - 12:49 PM)
Olafmol, if you want to automate things under windows, may I suggest you try batch files, using a command interpreter such as JPsoftware's 4DOS/4NT ?
sure..that's not really the problem for me
i first have to find/develop a useable approach.....
Olaf
I suggest you try the LAME "--scale" method first.
Encode 3 WAV files to MP3 using 3 slightly different scaling values (for example 0.90; 0.91; 0.92), then decode them and calculate each RMS value in Matlab (or maybe CoolEdit Pro). You should be able to distinguish between the 3 mp3 files, and they should sound identical of course.
Edit: When you know the difference is there, you can try programming the measuring tool in quickbasic or something
Encode 3 WAV files to MP3 using 3 slightly different scaling values (for example 0.90; 0.91; 0.92), then decode them and calculate each RMS value in Matlab (or maybe CoolEdit Pro). You should be able to distinguish between the 3 mp3 files, and they should sound identical of course.
Edit: When you know the difference is there, you can try programming the measuring tool in quickbasic or something
QUOTE (NumLOCK @ Feb 18 2003 - 01:03 PM)
I suggest you try the LAME "--scale" method first.
Encode 3 WAV files to MP3 using 3 slightly different scaling values (for example 0.90; 0.91; 0.92), then decode them and calculate each RMS value in Matlab (or maybe CoolEdit Pro). You should be able to distinguish between the 3 mp3 files, and they should sound identical of course.
Edit: When you know the difference is there, you can try programming the measuring tool in quickbasic or something
Encode 3 WAV files to MP3 using 3 slightly different scaling values (for example 0.90; 0.91; 0.92), then decode them and calculate each RMS value in Matlab (or maybe CoolEdit Pro). You should be able to distinguish between the 3 mp3 files, and they should sound identical of course.
Edit: When you know the difference is there, you can try programming the measuring tool in quickbasic or something
i will prolly give it a go, but it's not an ideal solution at all, as it cannot be performed automated in realtime.... it just takes to long and takes to much CPU performance to encode .WAV's on the fly after a purchase..
Olaf
To avoid multiple encoding there are two possibilities:
1. Take 1 mp3 and apply some changes
- The easiest way would be amplification=modifying gain for some frames - the problem is that to be lossless it shoud be a multipe of 1.5dB which could be audible
2. Take 2 (or more) mp3s, cut in several pieces and join to produce a unique pattern (1a-2a-3b-4a-5b-...)
2.a. If file b is produced by aplying mp3gain (amlification -1,5db) to file a, you can cut/join with any filesplitter without caring about mp3 structure - again 1,5 dB will probably be abxable sometimes.
2.b. If file b is produced another way, e.g. using lame 3.92 instead of 3.90.2 (Yes, this works - I tried it. There are differences in decoded wav files that would be enough.) or applying a very slight amplification (-0.01dB) before encoding, the cutting joining has to be done by special tools like mp3directcut. This is the best possibility IMO. There are many mp3-joining/cutting tools, but I don't know which would be best for "batching".
1. Take 1 mp3 and apply some changes
- The easiest way would be amplification=modifying gain for some frames - the problem is that to be lossless it shoud be a multipe of 1.5dB which could be audible
2. Take 2 (or more) mp3s, cut in several pieces and join to produce a unique pattern (1a-2a-3b-4a-5b-...)
2.a. If file b is produced by aplying mp3gain (amlification -1,5db) to file a, you can cut/join with any filesplitter without caring about mp3 structure - again 1,5 dB will probably be abxable sometimes.
2.b. If file b is produced another way, e.g. using lame 3.92 instead of 3.90.2 (Yes, this works - I tried it. There are differences in decoded wav files that would be enough.) or applying a very slight amplification (-0.01dB) before encoding, the cutting joining has to be done by special tools like mp3directcut. This is the best possibility IMO. There are many mp3-joining/cutting tools, but I don't know which would be best for "batching".
QUOTE (Artemis3 @ Feb 17 2003 - 11:05 PM)
There is no point in working against your consumers needs, open the doors, see the reality. Time has changed. Make sure your clients get the message too, they will anyway, either the easy or the hard way. You choose.
Will you treat your own customers as friends or enemies? Choose.
Will you lie to your client and sell them a system you know is going to be defeated?.
Will you treat your own customers as friends or enemies? Choose.
Will you lie to your client and sell them a system you know is going to be defeated?.
Man, You said everything...
I agree 100% with you.
Looks like "some"companies and associates, hate - or want to fool - their own costumers.
As expected, most ISP's just started selling music online also, clearly following all the "standards"(poorly encoded, non-burnable, time-bombed, expansive etc).
LIF
This is a "lo-fi" version of our main content. To view the full version with more information, formatting and images, please click here.