Discussion in 'Live Sound' started by Simmosonic, Apr 3, 2008.
I'm looking for an mp3 encoder suitable for acoustic music.
That is such a LAME request. (search LAME MP3 encoder)
Otherwise, Adobe Audition has the MP3 Pro encoder built-in to the software, which works quite nice.
I still use my old XING MP3 version 1.0 & 1.5 encoder that I got 10 years ago and still sounds perfectly adequate.
One of the things I've discovered regarding MP3 encoding is that too many recordings have a strong spike at 15kHz. This is frequently due to television horizontal sync circuitry within television sets, that manages to get into your audio equipment. The encoder sees this signal and devotes too much processing to try and retain it. You'll only know this if you can view your production with an audio frequency spectrum analyzer such as the one built into Sound Forage, Adobe Audition, Steinberg Wave blab. This is where a handy notch filter can help improve your MP3 encoding.
At 128 kilobits per second, response should only go out to 15kHz. But I believe that things still get screwed up because of this? But that I'm probably just hallucinating?
I don't want a gold record. I want titanium.
Ms. Remy Ann David
I usualy export MP3 (CBR 320kb/s) from the editing software (Sequioa), good enough for most clients that exept MP3..
The best stand alone software I've used (PC) is Illustrates dbPowerAMP, its a veritable "the swiss knife of audio processing"... (Well wort the registration fee!) - For MP3 is use LAME, as sugested by Remy above..
Most of my clients still prefer a lossless alternative, so I use the above to convert to FLAC..
Simmo, the encoder in Wavelab is LAME.
Thanks so far, Remy, ptr and Mr Spearritt.
I've been using the LAME encoder in Wavelab. I'm making mp3s of my field recordings to place as streaming audi files on my forthcoming 'blog matrix' (an interconnected set of audio-related blogs I'm working on at the moment).
I've tried a few of the encoders I have in Wavelab, including LAME and Fraunhofer. Some don't do such a good job, resulting in whistling artifacts (we used to call them 'birdies' back in the days when I had a CEDAR system). I find myself rendering the same file to a few different encoders, then listening back to see which one does the best job for the recording. But I was hoping someone would tell me that there's some superb new thing out there that creams everything else. Possibly not?
Not to my knowledge, all MP3 is very ugly. Here's a quote from Warren Ross (hope he doesn't mind me quoting him) on Glenn Meadows mastering board. It shows the lengths some will go to, to improve the result as much as possible.
When I get a request for MP3's of the mastered tracks (after bitching about it) I usually prepare the files first by Low-pass filtering the tracks around 13-16kHz and high-pass filtering at around 90-100hz with a high quality liner phase EQ, I use the Algorithmix Orange Lin EQ. MP3 encoders will usually high & low pass filter with sharp shitty sounding filters, so I use my own liner filters.
If it is hip hop and huge bass is wanted, I'll sometimes use Waves Maxx Bass to create some "fake" bass (it synthesizes harmonics and mixes them in with the original signal) to make up for the pass filtering. Then run the MP3 encoder at the highest bit rate the client can accept for his/her purpose.
My clients tell me these things make for a huge quality difference when listening to their tracks on simple laptop or home computer speakers. It is definitely a little time consuming to prepare a totally separate EDL for MP3's but the difference in quality is not subtle.
I have been getting close to taking similar measures, actually. I even labour over whether to encode the 24/96 file directly, or to SRC it to 44.1k and dither it 16-bit beforehand. In one example, I am sure the results were better after doing that.
What Warren says, and does, is very important for anyone hoping to impress others with their recordings - perhaps in the hope of generating some income. I have mastering engineer friends now who are making special MySpace versions of their client's mastered tracks, compensating for the shortcomings of MySpace's audio capabilities. To those clients, it is no longer good enough to make the recording sound as good as possible through reasonable speakers - there are entirely new stages in the signal path from performer to listener, and they must be factored into the mastering process.
From the not-ver-extensive testing I've done, Lame performed the best. If you are having artifact issues, I'd suggest VBR, perhaps with an average around 128 and see if that help. For my own listening, I use VBR and the EXTREME preset, which I think is an average of about 256.
If yu want to test/play with this, and that is a very good idea, download the free Swiss Army knife: foobar: http://foobar2000.org/
It is usually up to date on the LAME releases, or close enough. I would suggest using the VBR encoding technique as it is the most efficient and assigns the best parameters on an "as needed" basis. Set it up for ~192KBps and you should be good to go.
If you need any help, contact me and I will walk you through it: email@example.com.
Separate names with a comma.