Okay, this is probably a dumb question, I am new to the MP3 scene, but I understand how to rip CD's using EAC and then convert them, usually using LAME -vbr 5 for storage and playback on a car MP3 deck or computer drive.

I know this is low-quality, but I really can't hear the difference.

I have some MP3's that were ripped at much higher bitrates, which I do not have the original CD's for. Is there a way to use Lame to convert these using the original MP3's as source files.

(I guess I could burn them to .cda files and then rip, but that seems like an unnecessary step.

Thanks in advance!!!!

It isn't. A multiformat test carried out in 2005 revealed that the vast majority of listeners found ~128 kbit/s LAME very enjoyable up to transparent. A lot of audiophiles participated, since the test was organized by a member of HA.


LAME accepts MP3 as source data. If you don't want to use the command line encoder you could also go for foobar2000 as a graphical frontend. Its converter accepts all kinds of input codecs which can be handled by foobar and makes sure that the entire metadata is transferred to the resulting MP3.

You just have to set up the LAME command line once in order for the converter to use it: -S -V 5 --vbr-new - %d

This will make a good encoding preset. Note that the "--vbr-new" is obsolete in the current 3.98 Beta, but it's suggested for the 3.97 Release version.

Actually, you are correct there also. I actually encoded all the formats and couldn't even hear a difference down to V9 - I don't have good MP3 equipment at home so I was using a test CD on car stereo and bookshelf systems at a big-box store, but I figured a few extra MB's wouldn't hurt me, so I bumped it up to V5.

Okay, that's easy!!! (I've never used Foobar 2000 - I have used RazorLame) Foobar looks good, though. I am still using 3.97.

Thanks for the help!!!!

Okay, I’m ¾ of the way there . . .

I didn’t see options to do this with Foobar, but I am using RazorLame 1.1.5

I have two problems.

First I would like to simply copy the files to a new folder and then have LAME re-edit them in place. It gave me an error about the input and output file having the same name, so I had to create a \remix directory for it to write to. I can deal with that okay.

Second, though, is that it stripped the ID3 tags from the input files when it encoded them.

The command line I was using was:
-V 5 --vbr-new --lowpass 20

I tried with the command line I use with EAC (works fine there) which was:
-V 5 --vbr-new --lowpass 20 --add-id3v2 --pad-id3v2 --ta "%a" --tt "%t" --tl "%g" --ty "%y" --tn "%n" %s %d and it gave me an error message.

I also tried
-V 5 --vbr-new --lowpass 20 --add-id3v2 --pad-id3v2
and it stripped the tags out.

I learned all my EAC and LAME stuff from the Radified and The Coaster Factory, jthz.com, Chris Myden’s page and the sticky thread on here, so I’m pretty green at what all the -flags do.

Thanks again.

So far I haven't found any way to retain the ID3v2 tags with the command line encoder as well. LAME's documentation doesn't provide any infos concerning this issue, and I'm also lacking experience using it due to me having gone for the alternatives Vorbis & AAC long ago. But nonetheless, using two batches in conjunction with MP3tag I created a small workaround. This workaround will also make the separate /remix directory unnecessary, because it relies on temporary .mpx instead of .mp3 files.

Create a text file named e.g. MP3.cmd and enter this:


Don't forget to set the correct path to lame.exe in the "set encoder=..." line. Besides, this batch is based on the idea of someone else, but I don't remember the source anymore.

Another text file, named something like Del&Ren.cmd:


Put both of them into your audio collection's parent directory, e.g. C:\Audio. Be careful not to make a mistake here, since these scripts will recklessly process any MP3 data found in the directory they're placed in, including the subdirectories.

First of all, start the MP3.cmd script, it will recursively search for your MP3s and encode them to -V 5 equivalents with a temporary .mpx extension. I didn't include your lowpass option since there's no need for such an extreme lowpass, the default ~16 kHz one is just fine. As soon as the encoding process will have been completed, you'll have to load the parent directory into MP3tag. Make sure the application's allowed to both remove and write ID3v2 tags from/to MP3 files prior to processing anything. Then select the whole collection at once and choose the "cut tag" option in the right-click menu. Do not rearrange the files' order in MP3tag's list (e.g. by pressing F5 or clicking one of the columns) or even close the program down afterwards, otherwise the tagging will end in a catastrophe!

With the metadata having been copied to the clipboard, you can finally both delete the source MP3s and make the .mpx files take their place by running the Del&Ren.cmd batch command. A last step will be to return to the MP3tag window, select all files again and choose the "paste tag" option.

I'd suggest you to test the setup a bit, by copying one or two of your MP3 directories together with the two .cmd batches elsewhere and processing only these first. If everything worked out as intended you could go for the real thing afterwards.

This, of course, is only a workaround. I'd clearly suggest you to make yourself familiar with foobar's converter instead, it's way less painful since it doesn't include the issues related to transferring the metadata over to the new MP3s. In addition to this it's also able to conduct a ReplayGain scan.

Okay - First a couple of things just so I am on the right track . . .

If I used the /remix folder, I got the files into the format I wanted them. I suppose I could cut and paste with MP3Tag to get the tags over to the new files (I had never done that before).

Turns out this was harder than I expected.



Hold on, I'm using LAME 3.97. I read -lowpass 20 should be used as it filters out sounds between 20K and 22K. Are you saying LAME usually filters everything above 16Khz and this is fine. I thought FM audio was 50-15K and CD's were 20-20K, and FooBar's Equalizer has slides at 14K and 20K, so 16K sounds awfully low to me, but I'm a newb here. Please advise.


I don't mind using FooBar. I originally only saw playback options, but I looked again and saw that there is a right-click convert option.

Questions:

What is a ReplayGain scan and how do I do it? (Found that here: http://www.bobulous.org.uk/misc/Replay-Gain.html)

In Foobar, Under File - Preferences - Tools - Converter, it shows:

Name AVG BitRate Settings
MP3 (LAME) 190 kbps V2, Fast

Do I need to tell FooBar the path to LAME, and if so, where do I do that?

Do I need to set the custom settings, and can I just put in "-V 5 --vbr new" on that line?

The advantage of EAC was that it would read the file twice during the ripping and verify that it read it correctly. Does FooBar also do this, or is it not required with MP3 source files?

Thanks again!!!!

It's correct that a CD's frequency range theoretically reaches up to 22 kHz (half the sample rate of 44 kHz). And it's also correct that in theory the ear can hear up to around 20 kHz - at awfully loud levels. But in practice the human hearing's limitations already become obvious far below this value, high frequencies are very hard to impossibly to perceive in music, dependant on the listener and especially her/his age. Buzzy provided an interesting graph concerning this topic in another posting: http://www.hydrogenaudio.org/forums/index....st&p=492123.


ReplayGain is a technique to losslessly set the perceived average volume of your music to a reference value of around 89 dB. This is very useful since past recordings, expecially those from the early 90s and before, didn't have the doubtable pleasure to become victims of the so-called Loudness War. Recordings have been mastered at increasingly louder levels for some years now, resulting in decreased quality caused by clipping, as well as the listener being forced to utilize the volume knob every time the player switches between albums. This is especially pleasing if you were just listening to a well-mastered 1980's recording via your headphones and the player switches to an extremely loud, overcompressed 2007's one. Salutation to the eardrum!

ReplayGain is available in two versions: Track- and album-based. The former calculates an album's tracks independently from each other, the latter measures the whole album's average loudness and modifies its tracks' volume equally to each other. Album gain is the preferred one since it doesn't alter the intended dynamics between the songs. This is especially valid for live recordings and concept albums, where the tracks usually merge into each other. Quite annoying if the volume suddenly changes at the moment the player switches to the next track.

More about this topic: http://replaygain.org


The best possible option is to set up a custom encoder. Navigate to the converter (File --> Preferences --> Tools --> Converter) and click "Add New" next to the "Encoding Presets" list. The "Commandline Encoder Settings" will pop up, choose "Custom" as encoder there and point to lame.exe in the line below. Parameters should be something like -S -V 5 --vbr-new - %d, format is lossy, the rest is just descriptive.

You can now choose your new encoding preset in the "Converter Setup" by right-clicking the files you wanna encode to MP3 and choosing one of the "Convert..." options. It's suggested also to check the "ReplayGain-scan output files as albums" option there. This will result in foobar adding ReplayGain informations to the ID3v2 and, if existent, the APEv2 tags. ReplayGain tags are just added to the metadata, they don't manipulate the files themselves, making this change completely innocuous and reversible. Foobar can use the tags if you move to its playback options in the preferences and configure the ReplayGain section there. Source mode should be "album", processing in turn "apply gain and prevent clipping according to peak". You could also choose to use a preamp setting for RGed music in case the files were a little too quiet, though I wouldn't suggest to set it above +3 dB.


This is required for flawless CD ripping, not for encoding data that's already saved on the HD. There's so-called 2-pass encoding available for certain encoders using average bitrate (ABR) encoding mechanisms, but that's another thing influencing the resulting quality of the target file. It doesn't have anything to do with reading accuracy.

Thanks for that link to Buzzy's post. The thing I was missing was that LAME was already using a 16K LPF for the -V5 settings, so my 20K LPF was indeed meaningless.

Thanks also for explaining ReplayGain and walking me through FooBar. I am looking forward to re-working my files now.

Again, much thanks!

Where's the anti-transcoding horde? You're late. You should have been here hours ago. For shame.

Don't transcode!

so is this the same as lossy to lossy transcoding or different?

The same. As long as the source is lossy and the output is lossy, then the whole process is lossy transcoding.

but, is it uncompressing the mp3 and re-compressing it?
Is there the same quality loss? I'm just really confused.

The cabal has spoken, bah! (or is that baa?)

Let the guy ABX it for himself.

Yes and yes (well, the same quality loss as wave -> mp3 -> wave -> mp3)

As greynol stated, it is technically a different procedure but you will end up with the same quality loss. Going from wav to mp3 to wav to mp3 will result in the same quality loss if you just went from wav to mp3 to mp3. That extra step of going to the wav format doesn't add any quality what so ever.

For grins I went ahead and compared

wave -> mp3 -> mp3(2)

to

wave -> mp3 -> wave(2) -> mp3(2)'

So long as I made sure to pad wave(2) with 1105 samples of silence, mp3(2) and mp3(2)' were identical (bit for bit).


EDIT: BTW, all the encoding/decoding/trancoding were done with Lame 3.97 at the command line.

There is indeed no difference between going MP3 > WAV > MP3 and MP3 > MP3. Whatever you use to convert has to decode the MP3s anyway, you just don't see it happening.

But there is a difference, at least at the command line:

Lame doesn't ignore the first 1105 samples when going from mp3 to mp3 as it does when going from mp3 to wave.

Maybe the results are different when using foobar2000, since it provides for gapless decoding?

Wow, somehow I missed all the later replies to my post (no E-mail notification - or it was buried).

I can see how in theory lossy > lossy would be bad, but I don't think my ears are good enough to hear the difference. However, I do have some new questions.

Okay - For the most part it worked.

I couldn’t get Foobar to just straight convert the file to a lower bit-rate file in the same folder with the same name, it gave me a sharing violation. What I settled for was setting up a name format that was missing one space from the original, and then manually adding the space in later.

Also, my original file had both ID3v1 and ID3v2 tags. The converted Foobar file only seemed to have V1 tags if the V2’s weren’t different, but I was able to simply open MP3tagtools.exe, select all files, and select synchronize tags, so that is simple enough.

The Replay Gain function solved some mysteries, and created other questions.

First off, at home I can only play MP3’s on a pair of cheap PC speakers, but I am planning to get a car MP3 deck. I previously tried ripping a track from a CD at settings from V0 to V9, and all but V9 sounded transparent to me when I listened on display car decks and bookshelf speakers at a big box store.

So my ears are not that sensitive.

When I tried converting an MP3 source that was originally at CBR 224 to VBR 5, (both using RazorLame and Foobar), listening to the file in WMP 10 (10.0.0.3993), I thought the VBR 5 version sounded a bit less “crisp”. This didn’t make sense, as the CD rips were transparent on better equipment, and later things will contradict this. Maybe it was just mind games with my seeing the lower VBR and assuming it must sound worse.

Anyway, I re-did the Foobar version and then used Foobar to do a RG scan of a copy of the CBR 224 file. I then copied the original non-RG CBR224 file, the RG CBR224 File, and the RG VBR5 file to a directory. All three files now sounded transparent in WMP10 and Nero Showtime, however, the non-RG file sounded much better in Foobar. I checked and Foobar had reduced the file by 5 dB.

I am not sure I like the feature, and I verified that only Foobar seems to support it. I prefer using WMP to play files and I verified that none of the car decks that I am considering support RG, so I am not sure it matters, but I am now wondering:

Is 89 dB a “magic” number? Could I set RG to 94 dB? This would make less of a change from my CBR224 source, although I realize other source files might still be affected.

There is an option to apply RG to the files rather than just the tags. This would be recognized by my car CD player. Is this worth considering? 95% of the files are either ripped from CD or transcoded from a higher bitrate original that I am keeping, so if I didn’t like it, I could simply re-rip the files, although that could be annoying to have to do for a large number of files.

Thanks in advance!!!

Click to view attachment or mp3gain or search HA / wiki / FAQ.

Cheers,
David.

Thanks - MP3gain looks cool. I think I'll play with it some.

Found some very interesting results with MP3Gain.

Doing Album Analysis, most of the tracks that I ripped from CD had only one or two tracks with clipping. Most of the albums that were I did not rip myself had clipping on all EXCEPT one or two tracks. (Many of these were re-masters, though, so that might be why. Many of them also had volume settings in the 93 97 dB range).

I again tried transcoding the files from cbr 224 to V 5, this time using Foobar without RG enabled. I then copied all the files and ran MP3Gain on them.

The adjustments were the same and the cbr 225 and V5 non-RG and with-RG files sounded transparent to me.

The -RG files were noticeably quieter (in WMP), and to me sounded a bit “less crisp”, although I didn’t play with the volume enough to see if playing the RG files louder would give them the same “crispness” as the non-RG files at lower volume.

I guess what I have questions about at this point:

I understand clipping is bad, but how come so many of the files indicated that they were in the clipping range, but sounded fine to me. Also, how bad is clipping really, in terms of this thread: http://www.hydrogenaudio.org/forums/index....showtopic=55363 where several users have said they couldn’t even hear it?

I like the idea of all albums being at a common volume, however, I can see the following drawbacks to the 89 dB volume level, and would like comments on this:

PC I like the MP3’s at moderate volume, but don’t want the beeps and startup .wav’s really loud, and I would prefer not to have to adjust my speaker volume everytime I switch sources.

Car I don’t really want to want to have to turn the volume DOWN when going from CD to radio. Also, I will be using the factory low powered speakers initially I don’t know if having the volume turned way up with the sound volume moderate means the amp is working harder and the speakers are taking more power even though the sound is not as loud.

Portable Basically same as above, does not having the heard volume high but having the volume knob way up increase battery drain?

Thanks for all advice/suggestions!!!

1) If you take a loud modern recording and mp3 it, you end up with an mp3 that clips, but the loudness is the same as the original.

2) If you take an older or quieter recording, mp3 it, and raise the volume, you will also end up with an mp3 that clips. However, this clipping will be dramatically more audible than that in (1). Depending on genre, you can find examples that sound truly awful.


Playing "quiet" (RG) tracks with the amplifier volume set high, or "loud" (non-RG) tracks with the amplifier volume set low doesn't seem to make a dramatic difference to power consumption. It's when you can't get the amplifier volume high enough, or the amplifier noise increases audibly, or the volume becomes dramatically different from another source (as you've suggested) that problems arise.

If the RG'd files sound "less crisp" it's 100% due to the volume difference. There is no quality difference. In some cases your audio equipment (sound card/amp/speaker) might be audibly different with quieter audio (so you could say there is a quality difference). In most cases it's just and artefact of human perception - just turn the volume up to compensate, and it will sound the same as it did.

There's nothing wrong with using 93dB as an RG target, except where this causes existing tracks to be increased in volume and to clip in the process.

Hope this helps.

Cheers,
David.

Okay that makes sense.



Okay, this is also what I expected. Let me add that on the PC I am using on-board sound card and very cheap speakers, and I'm not really wanting to turn the speakers up 3/4 of the way to hear MP3's and then down to 1/4 of the way for gaming, etc. (Of course, the MP3's will be mainly for the car deck once I get that.)

Okay, next dumb newbie question - if my original is a modern re-mastered file at 93dB and MP3Gain says it is clipping but this is acceptable, how do I know that an older file increased to 93dB when MP3Gain says it is clipping is not acceptable. IOW, in both cases, MP3Gain just puts a Y in the clipping column, so how do I know when this is okay and when it is not okay.

Ask yourself, do you really need all of those beeps & waves? On my PC, I find them annoying and have them ALL disabled. However, if there are some that you truly can't live without, then you could always substitute quieter versions for the originals. Just a thought.

I did that (though I just switched them all off later!). It's easy enough to load the .wavs into an editor, or even run them through wavegain!




That's easy enough. If mp3gain doesn't change the gain, or turns it down, and still reports clipping - then that's likely to be the less audible kind.

If mp3gain increases the gain, and then (and only then) reports clipping, then that's likely to be the very audible kind.

mp3gain shows you the gain correction it will apply in the appropriate columns ("track gain" and "album gain"). If you push the target high, and want to ensure quieter albums don't clip, don't process them if the gain is positive and the clipping says "Y".

However, if this is the method you chose, you might as well not bother using mp3gain or ReplayGain at all. Apart from taming some very loud tracks, it won't do much.

As the FAQ says, it can't work properly if you won't turn the louder tracks down but can't turn the quieter tracks up!

http://replaygain.hydrogenaudio.org/faq_quiet.html

Cheers,
David.

I hadn't thought about reducing the .wav file volume for the intros and beeps. However, I also have a lot of games that I would like to have quieter than my CD's, and I'm not about to hunt down and reduce the gain on all those. Likely I'll just get used to using the volume control.

Okay, I think I have this down, now, but if you would be kind enough to look at the attached screenshot and confirm.



This album was originally at about 93db, and all but the last track indicate clipping (but it is likely okay). Originally, MP3Gain had a target value of 89 and showed about a 6db cut on each track and the clip(track) (4th column) was empty. As you can see, I set a target value of 94db, and the last track now gets a 3db boost, and shows clipping, which is likely bad. The 5th and 6th tracks get a 1.5 db reduction and no longer clip. Tracks 2, 3, 7, and 8, get no gain adjustment, and still clip, but this is likely acceptable.

Question - Tracks 1 and 4 were clipping originally and now get a 1.5 dB gain and still show clipping. I am guessing this is now not the okay clipping, but likely the "bad" clipping since the gain was increased. Correct?

As I see it, my options are as follows:

Leave target volume at 89db, this will eliminate clipping on most all the albums, but they will be fairly quiet, and I will need to rely on my PC or car amp to get the volume I desire out of them.

Set target volume at 94. This will make all albums pretty loud, but in some cases, I will introduce "bad" clipping, like above. In this case, I could reduce the target volume until the 4th column no longer showed "Y", but this would make this album quieter than others that did not clip at 94db, and thus reduce the intent of MP3Gain normalizing all CD volumes.

Target Volumes between 94 and 89 would be a trade-off between how loud I can set my average volume, and how uniform the volume would be. IOW, higher target volumes mean overall all albums are louder, but there is more variation in volume, b/c some albums won't be able to reach the target without clipping, so their target value will be lower. Lower target values will mean more albums can reach the target without clipping, but the overall volume will be lower.

What I might need to do is test record a track with the gain at 89, 89.5, 90, 90.5 . . . 93, and then test this on a car deck and see when the CD volume is closest to the FM volume, and try to set this as a target. (And/or pick a target and then see how many times I have to reduce it and to what level, and then let that be my new target. For example, the example above shows no clipping at 91.6, but does at 91.7, so if the majority of my albums could handle 91dB, that would be a "safe" or reasonable target.

Did I blatantly miss anything?

Did some comparative testing and thought I would post the results.

I took four samples of the same track.

One with MP3Gain volume set at 89.6 (Target 89.0), one at 91.1, one at 92.7, and the unmodified original at 94.2. Burned them all to a test disk and tried them on several car stereos in the Best Buy display.

The 89.6 sample was louder than FM on all the stereos, sounded clean (no amp hiss, or high volume required), and is what I will stay with. The other samples were louder, but also seemed a bit compressed, although that could be because of other factors beyond the actual MP3Gain settings.