trip_likeido
That is such a coincidence !!!!!!!!!!!!
I have been doing google searches for cheap online music stores that carry CD singles. The problem is that many Euro/Trance MP3's are available on VINYL only !!! So, just like you said ... you either buy TT's and records to go along with it ... or just use the terribly-ripped mp3's floating around.
Popular stuff you can find cheap single CD's for. I bought Janny from the Block, Green velvet and a whole wack of other singles for $4 CDN each.
I check ebay every so often. Some good cheap stuff on there.
I have been doing google searches for cheap online music stores that carry CD singles. The problem is that many Euro/Trance MP3's are available on VINYL only !!! So, just like you said ... you either buy TT's and records to go along with it ... or just use the terribly-ripped mp3's floating around.
Popular stuff you can find cheap single CD's for. I bought Janny from the Block, Green velvet and a whole wack of other singles for $4 CDN each.
I check ebay every so often. Some good cheap stuff on there.
Posted Thu 23 Jan 03 @ 4:50 am
jackcu
I know what you mean guys - I'm luckier in so far as my customers only want to hear the latest chart stuff and that's easily available on CD. The major problem here in the UK is the stupidity of the "chart race".
When I was a kid, a single would be played on the radio for about one week maximum before being released in the shops. I remember buying T-Rex's "Solid Gold Easy Action" (giving away my age) and it going straight in at #1. That took the best part of 1m sales to achieve.
Nowadays, the songs are played on the radio and TV for up to 4-6 weeks before they are released in a cynical attempt to get every song to the top of the charts on its first week. Even then, they can hit the top spot with extremely low sales in comparison to 10 years ago. The trouble is that I get asked for this stuff and I can't buy it anywhere unless it's on an album.
Damn those record companies ;)
Cheers,
Jack.
When I was a kid, a single would be played on the radio for about one week maximum before being released in the shops. I remember buying T-Rex's "Solid Gold Easy Action" (giving away my age) and it going straight in at #1. That took the best part of 1m sales to achieve.
Nowadays, the songs are played on the radio and TV for up to 4-6 weeks before they are released in a cynical attempt to get every song to the top of the charts on its first week. Even then, they can hit the top spot with extremely low sales in comparison to 10 years ago. The trouble is that I get asked for this stuff and I can't buy it anywhere unless it's on an album.
Damn those record companies ;)
Cheers,
Jack.
Posted Thu 23 Jan 03 @ 12:52 pm
Grimm
MP3 is a lossy file, but depending on how it was encoded, what encoder was used, and what decoder was used can affect the quality of the output as well as hardware used.
For a comparison of a wav file and a 320kbps mp3 file goto my page here:
http://roadriders.no-ip.org/4images
I also have pictures of the sound quality that Atomix outputs. So if your unsure about what other may be speaking about when they say it sounds bad when the pitch is changed, you'll see it there.
Grimm
For a comparison of a wav file and a 320kbps mp3 file goto my page here:
http://roadriders.no-ip.org/4images
I also have pictures of the sound quality that Atomix outputs. So if your unsure about what other may be speaking about when they say it sounds bad when the pitch is changed, you'll see it there.
Grimm
Posted Sat 25 Jan 03 @ 12:08 pm
jackcu
Hi Grimm,
Interesting test - however, I can't see the differences. Maybe I'm just being thick, but your graph comparing the original reference signal against the LAME endoded MP3 signal just looks like the same graph. As far as I am aware, MP3 has a brickwall at 14Khz - I've just analysed an MP3 and it looks more like 16Khz to me - but on your graph it is showing frequencies that are far higher (and the same as the reference signal). The filename in the picture is "frequencysweep.wav" whereas I would have expected it to be "frequencysweep.mp3" - just a thought.
Could you take a look and tell me what I'm not seeing?
Thanks,
Jack.
Interesting test - however, I can't see the differences. Maybe I'm just being thick, but your graph comparing the original reference signal against the LAME endoded MP3 signal just looks like the same graph. As far as I am aware, MP3 has a brickwall at 14Khz - I've just analysed an MP3 and it looks more like 16Khz to me - but on your graph it is showing frequencies that are far higher (and the same as the reference signal). The filename in the picture is "frequencysweep.wav" whereas I would have expected it to be "frequencysweep.mp3" - just a thought.
Could you take a look and tell me what I'm not seeing?
Thanks,
Jack.
Posted Mon 27 Jan 03 @ 4:04 pm
Grimm
Your seeing it correctly. The brick wall your referring to is most likely a poorly encoded mp3 file. Certain codecs (older Xing codec for example) will have a cutoff frequency of 16kHz, no matter what settings you use. Older versions of the Fraunhofer codec also had a cutoff of 16kHz when certain bitrates were used. But even some of the newer version still have the built in 16kHz cutoff unless you disable it or encode the mp3 file in a certain manner (ie, high bitrate, forced simple stereo, no vbr, who knows, I dont use any other codec cept Lame).
The codec I used to encode the wav was Lame 3.92MMX, with no cutoff frequency. The exact line used to encode the wav was: -b 320 -m j -h -q 0
As you saw, the Lame encoder does a very good job, decoding as well. There is a very small difference between the original wav and the Lame decoded wav. If you look closely in the lower frequency range, you'll notice there was a very small loss of frequency at one point, about .3db if I remember correctly. That is something not really noticeable by the human ear, no matter how good the equipment or ear is.
For a little more information on what I did, check out this thread.
http://www.atomixmp3.com/forum/display.html?topic=9149
Grimm
The codec I used to encode the wav was Lame 3.92MMX, with no cutoff frequency. The exact line used to encode the wav was: -b 320 -m j -h -q 0
As you saw, the Lame encoder does a very good job, decoding as well. There is a very small difference between the original wav and the Lame decoded wav. If you look closely in the lower frequency range, you'll notice there was a very small loss of frequency at one point, about .3db if I remember correctly. That is something not really noticeable by the human ear, no matter how good the equipment or ear is.
For a little more information on what I did, check out this thread.
http://www.atomixmp3.com/forum/display.html?topic=9149
Grimm
Posted Mon 27 Jan 03 @ 10:21 pm
Salva
Dump that gain thing and implant a good DSP pugin like Audiostocker Pro!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
Posted Tue 28 Jan 03 @ 11:40 am
trip_likeido
"How dynamic is your audio?"
Posted Tue 28 Jan 03 @ 5:02 pm
Liamalone
bit of a bugger to read this thread so far down the line.....too damn long!!!!
I just mess about with Atomix for pleasure and had no problem with the sound as I don't ask it to do a while lot........but
A couple of months agoI lined up a ripped wav file and the resultant mp3 file that I made - at 160kbps cos everyone argues 128 or 192.....I compromised - and when I played them in sync I noticed some difference in sound. I thought the mp3 file sounded better!!!! lol
Pity the guy playing on or before someone with good vinyl........then the difference would be audible.
later,
Liam
I just mess about with Atomix for pleasure and had no problem with the sound as I don't ask it to do a while lot........but
A couple of months agoI lined up a ripped wav file and the resultant mp3 file that I made - at 160kbps cos everyone argues 128 or 192.....I compromised - and when I played them in sync I noticed some difference in sound. I thought the mp3 file sounded better!!!! lol
Pity the guy playing on or before someone with good vinyl........then the difference would be audible.
later,
Liam
Posted Tue 28 Jan 03 @ 5:55 pm
jackcu
Hi Grimm,
Thanks for the reply - interesting analysis. However, I'm still confused. I have 2 questions:
1. What is the scale along the bottom of your graph for the reference sweep? You state that the tone goes from 20Hz to 20Khz (makes sense to me, being the normal range of very good human hearing). However, your scale runs from 200000 to 3000000 - what is this?
2. How big is the MP3 file of the tone? If is doesn't lose any frequencies over the original, i would expect it to be very large.
Many thanks,
Jack.
Thanks for the reply - interesting analysis. However, I'm still confused. I have 2 questions:
1. What is the scale along the bottom of your graph for the reference sweep? You state that the tone goes from 20Hz to 20Khz (makes sense to me, being the normal range of very good human hearing). However, your scale runs from 200000 to 3000000 - what is this?
2. How big is the MP3 file of the tone? If is doesn't lose any frequencies over the original, i would expect it to be very large.
Many thanks,
Jack.
Posted Wed 29 Jan 03 @ 2:06 pm
jackcu
Hi Grimm,
I've played your MP3 sweep in another spectrum analyser and get the same result as you do - which puzzles me. However, on thinking about it, maybe it's not so surprising.
As the sweep moves up, at any one point it is only playing one, single frequency. Therefore, the MP3 encoding process can throw away every other frequency at that point. This would result in the graph of the .wav and the graph of the MP3 being the same. However, music doesn't consist of one frequency at a time. If you were to encode a typical piece of music, the only way to compress would be to lose some of the frequencies or other audio information - at least that's what I think LOL.
Would you mind encoding a "typical" song for us and posting the results? If I'm correct the results will be rather different.
I'm interested in the fact that you have been able to encode without hitting a brickwall. I was under the impression that ALL encoders do this. I too use LAME and I'm getting a brickwall around 16Khz, as I mentioned. However, I'm running it through dbPoweramp rather than from a command line. I'll try your string and see what I get.
Many thanks again,
Jack.
I've played your MP3 sweep in another spectrum analyser and get the same result as you do - which puzzles me. However, on thinking about it, maybe it's not so surprising.
As the sweep moves up, at any one point it is only playing one, single frequency. Therefore, the MP3 encoding process can throw away every other frequency at that point. This would result in the graph of the .wav and the graph of the MP3 being the same. However, music doesn't consist of one frequency at a time. If you were to encode a typical piece of music, the only way to compress would be to lose some of the frequencies or other audio information - at least that's what I think LOL.
Would you mind encoding a "typical" song for us and posting the results? If I'm correct the results will be rather different.
I'm interested in the fact that you have been able to encode without hitting a brickwall. I was under the impression that ALL encoders do this. I too use LAME and I'm getting a brickwall around 16Khz, as I mentioned. However, I'm running it through dbPoweramp rather than from a command line. I'll try your string and see what I get.
Many thanks again,
Jack.
Posted Wed 29 Jan 03 @ 2:26 pm
MP3 just takes away souns which cannot be heared and are "behind" the ones we *can* hear. So it is possible the "forms" of the original .wav file are the same as the one of the .mp3 file. I guess.
Roderick
Roderick
Posted Wed 29 Jan 03 @ 4:15 pm
Grimm
Thats exactly why I choose a frequency sweep to compare the outputs. Its a simple audio file to encode that would have the full range of audiable sound, from 20Hz, to 20kHz. Niether the encoder nor decoders sould have any trouble reproducing the audio accurately, and the resulting comparison graphs wouldn't be anything complicated to look at. Any difference on such a simple file by the decoders would show thier shortcomings.
On a regular musical track, the difference seen between the original track and encoded track would not only be the fault of the decoder, but also the encoder throwing away some information. I wasn't making this a comparison of the encoder used, but a comparison of the decoder's/mp3 player's output. The simple frequency sweep would show any problems that the decoder has in playing back the file, not of the encoder having problems encoding the file.
And check the help files with dbPoweramp. They may have added in a extra "feature". If it uses the regular Lame.dll file, then it sould be something easy to change or disable. If it has the Lame codec built in, it may be something hard coded into the program. If thats the case, I say dump it. Get the regular Lame win32 distribution with has both the command line .exe and a windows .dll file and use the RazorLame frontend.
Personally, I encode all my files with a 19kHz low pass filter, since the average adult listner's hearing is around 18kHz. Why the norm is considered 16kHz, I'll never know.
Grimm
On a regular musical track, the difference seen between the original track and encoded track would not only be the fault of the decoder, but also the encoder throwing away some information. I wasn't making this a comparison of the encoder used, but a comparison of the decoder's/mp3 player's output. The simple frequency sweep would show any problems that the decoder has in playing back the file, not of the encoder having problems encoding the file.
And check the help files with dbPoweramp. They may have added in a extra "feature". If it uses the regular Lame.dll file, then it sould be something easy to change or disable. If it has the Lame codec built in, it may be something hard coded into the program. If thats the case, I say dump it. Get the regular Lame win32 distribution with has both the command line .exe and a windows .dll file and use the RazorLame frontend.
Personally, I encode all my files with a 19kHz low pass filter, since the average adult listner's hearing is around 18kHz. Why the norm is considered 16kHz, I'll never know.
Grimm
Posted Thu 30 Jan 03 @ 1:15 am
jackcu
Hi Grimm,
I downloaded Lame 3.93 and tried running it from the command line with the "--R3mix" switch and have to say I'm impressed with the quality. Can you suggest any other standard settings that produce the best results?
I played a few tracks on my monitors at home last night and I could still hear the difference between the .wav and the MP3. However, it was much, much less noticable than before and I'm beginning to think it might be the MP3 player that is colouring the sound. I have tried it through Sonique and the WMP and it sounds different on each one.
The funny thing is that played through AtomixMP3, they sound the same. I read your analysis and saw that you reckon AtomixMP3 isn't as good at playing back the sound as some other players. It must also colour the sound when playing .wav files, 'cos I couldn't tell the difference.
Looks like your suggestion of the developers spending some time improving the quality of playback is a good one.
Thanks for the info and the feedback - you've changed my point of view on what is possible with MP3.
Cheers,
Jack.
I downloaded Lame 3.93 and tried running it from the command line with the "--R3mix" switch and have to say I'm impressed with the quality. Can you suggest any other standard settings that produce the best results?
I played a few tracks on my monitors at home last night and I could still hear the difference between the .wav and the MP3. However, it was much, much less noticable than before and I'm beginning to think it might be the MP3 player that is colouring the sound. I have tried it through Sonique and the WMP and it sounds different on each one.
The funny thing is that played through AtomixMP3, they sound the same. I read your analysis and saw that you reckon AtomixMP3 isn't as good at playing back the sound as some other players. It must also colour the sound when playing .wav files, 'cos I couldn't tell the difference.
Looks like your suggestion of the developers spending some time improving the quality of playback is a good one.
Thanks for the info and the feedback - you've changed my point of view on what is possible with MP3.
Cheers,
Jack.
Posted Thu 30 Jan 03 @ 1:25 pm
Well, why dont you buy a soundcard that enhance the quality of mp3???? that's way when you use Atomixmp3, the quality is better than some crappy soundcard in your system
Posted Fri 31 Jan 03 @ 2:00 am
jackcu
The soundcard has nothing to do with what I'm hearing. Grimm did the analysis on the actual waveforms - his work suggests that AtomixMP3's sound is not quite as good as some other programs, no matter which soundcard you use. I can hear a comparative difference using one card - it would still be there with a different card even if the overall sound quality improved.
Anyway I don't have a "crappy" sound card on my system :)
BTW, I'm not having a go at AtomixMP3. I LOVE this program and use it all the time. I'm just hoping that if there is some work to be done on sound quality that it will get on to the developers list.
Cheers,
Jack.
Anyway I don't have a "crappy" sound card on my system :)
BTW, I'm not having a go at AtomixMP3. I LOVE this program and use it all the time. I'm just hoping that if there is some work to be done on sound quality that it will get on to the developers list.
Cheers,
Jack.
Posted Fri 31 Jan 03 @ 3:29 pm
Yoda
I also agree that the sound quality of Atomix is not everything I'd like it to be. In fact, it's the worst feature of Atomix and I think it really lets down what is otherwise excellent s/w. I do my best to work around the problem by trying to keep pitch adjustments as close to normal as possible.
I'd like to hear the developers comment on what has been said in this thread, since it's been going for so long. Is there any acknowledgement that the quality is poor in comparison to other 'mp3 DJing s/w' and are there any plans to address this problem in future releases?
I'd like to hear the developers comment on what has been said in this thread, since it's been going for so long. Is there any acknowledgement that the quality is poor in comparison to other 'mp3 DJing s/w' and are there any plans to address this problem in future releases?
Posted Fri 31 Jan 03 @ 9:35 pm
trip_likeido
I noticed that the devlopers have not participated in such discussion. I don't blame them for it. It's like a Cell phone company admitting their coverage is bad.
I find that the quality gets Jeopardized regardless at what pitch setting you have it at.
I find that the quality gets Jeopardized regardless at what pitch setting you have it at.
Posted Sat 01 Feb 03 @ 3:42 am
jackcu
Good points,
Is there a way to get the developers involved?
I'm no expert on the way that the software converts the MP3's back to sound - is it as simple as using a different codec?
Grimm - you seem to understand this stuff - do you have a suggestion as to how the sound quality of the playback could be improved?
Cheers,
Jack.
Is there a way to get the developers involved?
I'm no expert on the way that the software converts the MP3's back to sound - is it as simple as using a different codec?
Grimm - you seem to understand this stuff - do you have a suggestion as to how the sound quality of the playback could be improved?
Cheers,
Jack.
Posted Mon 03 Feb 03 @ 3:11 pm
trip_likeido
These are top factors that contribute to bad sound quality
* The MP3 Files themselves; Low Bitates; poorly encoded, encoded from vinyl; Encoded in 480000 KHz
* Although most 16 Bit soundcards are of comparable quality, cards like the SB Live! the Terratec are the higher ends which should be used
* Atomix Gain too high
But alot of the time (for me) is when the pitch is adjusted, thats what degrades the quality. I'm pretty sure it's not he XAudio Codec. It is the pitch algorithm. Mixvibes must have a similar alogorithm too, as that program has the same effect
You can always enable the time stretching, that what you don't get that annoying hiss, however you get that 'gargled' music.
* The MP3 Files themselves; Low Bitates; poorly encoded, encoded from vinyl; Encoded in 480000 KHz
* Although most 16 Bit soundcards are of comparable quality, cards like the SB Live! the Terratec are the higher ends which should be used
* Atomix Gain too high
But alot of the time (for me) is when the pitch is adjusted, thats what degrades the quality. I'm pretty sure it's not he XAudio Codec. It is the pitch algorithm. Mixvibes must have a similar alogorithm too, as that program has the same effect
You can always enable the time stretching, that what you don't get that annoying hiss, however you get that 'gargled' music.
Posted Mon 03 Feb 03 @ 4:44 pm
Thibaud Dafaux
Ok, this thread is getting reaaaaaally long but I haven't seen any comments of the Atomix Team.
I would like to know if they intend to improve the quality or not. (I don't mean that in a rude way)
GnoedeL
I would like to know if they intend to improve the quality or not. (I don't mean that in a rude way)
GnoedeL
Posted Mon 03 Feb 03 @ 5:39 pm
