resampling mp3s?

[source]

Is there amy programs for the amiga that would take a 320 encoded mp3 and convert down to 128?
Thanks

[Hewitson]

There's a port of LAME, but I hate to think how long it would take to do the job. Why would you want to convert a good quality MP3 to a poor quality one anyway?

[pandy71]

At first try to run this: http://wiki.hydrogenaudio.org/index.php?title=MP3packer perhaps 320kps files are not 320kbps...

And same as Hewitson - i see no point to transcode mp3 files on Amiga where PC will be faster 10 - 100 times.

[source]

I just want to convert some of my high quality mp3s so that it will play on my MAS player for the amiga.

[Thorham]

This may seem obvious, but:

If you have the music available in FLAC/WAV or on CD use that instead. Recoding lossy formats is always a bad idea, because the losses are added up together. Always encode from non-lossy sources.

[pandy71]

Originally Posted by source

I just want to convert some of my high quality mp3s so that it will play on my MAS player for the amiga.

Go for PC and lame then..., if MAS deal with VBR go for it - this is way better than CBR, filter everything above 16kHz - MP3 has "flaw" by design and usually this part of spectrum not exist - but if something exist this is not even close to music thus for lower bitrate mp3 it is better to remove any signal from this part (not loos data bits).

[source]

Thanks for the input. I am using the 320 versions on my A2000 with a delfina. Over clocking the Delifna allows for it to play without stuttering. I don't know how to overclock the MAS player if can at all.

[pandy71]

Originally Posted by source

Thanks for the input. I am using the 320 versions on my A2000 with a delfina. Over clocking the Delifna allows for it to play without stuttering. I don't know how to overclock the MAS player if can at all.

Delfina is equipped with general DSP processor and clock speed directly affect processing speed - MAS3507 is dedicated DSP only for MP3 and it should play all bitrates up to 320kbps without any problem.

I would say that 320kbps is a bit overkill for usually mp3 files and also that usually mp3 with such high bitrate will be outperformed by mp2 (mp2 offer higher audio quality than mp3 usually from 224 - 256kbps).

Perhaps stuttering problem is not related to speed but insufficient bitstream parsing speed by CPU (not sure about MAS3507 but AFAIR it doesn't support elementary audio stream and need additional CPU processing - as opposite to for example STA013 where you can play mp3 files directly even from serial port - if serial port is fast enough - should be in Amiga case).

[Daedalus]

The MASPlayer uses its own crystal at 14-ish MHz, but I don't think overclocking it will do any good - the DAC also uses this signal for its timing. The limiting factor in our case however is the throughput from the parallel port - effectively the parallel port is emulating an I2C bus with software doing all the bus control signals. Therefore, a faster CPU in your Amiga will help, but also better coding. I'm not a particularly good coder, and the routine I wrote using C is able to decode 160 or usually 192kbps without stutter, but above that causes issues. That's on an '030/50.

The MAS Player application is written in assembler however and I seem to remember that coping perfectly well with much higher bitrates.

[john1979]

Originally Posted by pandy71

usually mp3 with such high bitrate will be outperformed by mp2 (mp2 offer higher audio quality than mp3 usually from 224 - 256kbps).

That's not true. MP3 will be transparent to the source material (sound the same as) for nearly all people at those bit rates. MP2 offers nothing above MP3 at any bit rate..

[pandy71]

Originally Posted by john1979

That's not true. MP3 will be transparent to the source material (sound the same as) for nearly all people at those bit rates. MP2 offers nothing above MP3 at any bit rate..

Common misconception - MP3 is lossy compression and it is always in objective term non-transparent - it can be subjectively transparent but for various people level of transparency can be different.
Side to this MP2 is free from smearing and pre-echo where MP3 by design can't be free from this. Most audiophile friendly lossy codec (Musepack - http://en.wikipedia.org/wiki/Musepack) is based on MP2 not on MP3.

Very fast for you:
http://en.wikipedia.org/wiki/MP2_%28...2_format_works

MP3 provide higher spectra resolution at a cost of temporal resolution, MP2 is particulary good for rapid transient sounds where usually MP3 produce noticable (sometimes even for high bitrates) distortions - complex audio scene is usually reproduced worse by MP3 http://www.hydrogenaudio.org/forums/...howtopic=74100

This is quite difficult to describe but perhaps you are not sensitive for this (personally i hate such warbling audio tone) - i observe that lot of people especially younger don't hear MP3 artifacts but this can be explained by fact that they have no experience with life instruments and/or non compressed audio and brain (perception) is adopted to mask this problem.

So MP3 with 128kbps will sound probably better than MP2 but MP2 with 320 - 384kbps will be way better than MP3.

btw check huge band above 16kHz in MP3 - it almost not exist unless energy in this part of spectrum is high enough to be encoded - MP3 intentionally save bits on this part of spectrum where MP2 is way more transparent...

[john1979]

Originally Posted by pandy71

Common misconception - MP3 is lossy compression

Of course I did not confuse lossless with lossy. Why did you think I did?

and it is always in objective term non-transparent

Means nothing. Either you can tell an encoding from the original through a double blind test (it's not transparent) or you can't (it's transparent to you). ABX testing takes away subjectivity and makes the test objective. Of course it does not matter what other people can hear, only you

Side to this MP2 is free from smearing and pre-echo where MP3 by design can't be free from this.

MP2 performs better on sharp transients it has to be said. However why I'd want to choose a codec that needs high bit rates to remain indistinguishable to the source is another question.

When we are talking about bitrates 256k+ I would still expect LAME MP3 to have less problem samples than MP2 anyway. LAME is highly tuned and has an excellent psycho-accoustic model.

So MP3 with 128kbps will sound probably better than MP2 but MP2 with 320 - 384kbps will be way better than MP3.

At high bitrates it most likely both would be transparent to the source. There would be no "better" codec.

As for Frank Klemm's baby, Musepack. It is surely a nice codec. I prefer aotuv vorbis myself. To me it is transparent at Q5 (~160kpbs) even on the worst problem samples.

btw check huge band above 16kHz in MP3 - it almost not exist unless energy in this part of spectrum is high enough to be encoded - MP3 intentionally save bits on this part of spectrum where MP2 is way more transparent...

A common mistake, spectral graphs mean nothing when it comes to audio encoding transparency. Only your ears and ABX mean anything. Surely if you paste links to Hydrogenaudio you should know that already?

[pandy71]

Originally Posted by john1979

Of course I did not confuse lossless with lossy. Why did you think I did?

Im not confused - i trying to say that with lossy compression we can't talk about universal - objective transparency - each person has own threshold of transparency.

Originally Posted by john1979

Means nothing. Either you can tell an encoding from the original through a double blind test (it's not transparent) or you can't (it's transparent to you). ABX testing takes away subjectivity and makes the test objective. Of course it does not matter what other people can hear, only you

Transparent for you doesn't mean transparent for someone else.

Originally Posted by john1979

MP2 performs better on sharp transients it has to be said. However why I'd want to choose a codec that needs high bit rates to remain indistinguishable to the source is another question.

Simpler compression model or you don't need to squeeze every possible bit to reduce required bandwidth - every lossy compression is compromise...

Originally Posted by john1979

When we are talking about bitrates 256k+ I would still expect LAME MP3 to have less problem samples than MP2 anyway. LAME is highly tuned and has an excellent psycho-accoustic model.

Isn't similar PSA is implemented in twolame encoder? Once again MDCT concept have own pross (better squize data) and cons (mentioned earlier).

Originally Posted by john1979

At high bitrates it most likely both would be transparent to the source. There would be no "better" codec.

I would say if you going for 320kbps then why you want to squeeze any possible bit?

Originally Posted by john1979

As for Frank Klemm's baby, Musepack. It is surely a nice codec. I prefer aotuv vorbis myself. To me it is transparent at Q5 (~160kpbs) even on the worst problem samples.

Fair, for me vorbis sounds OK even at lower datarates especially when compared to MP3 (as MP2 is not suitable for low bitrates).

Originally Posted by john1979

A common mistake, spectral graphs mean nothing when it comes to audio encoding transparency. Only your ears and ABX mean anything. Surely if you paste links to Hydrogenaudio you should know that already?

Yes and know but this is additional discussion - this was only point that for people able to hear over 16kHz, MP3 can be insufficient due design limitations - personally i am old enough to not hear above 17 - 18kHz thus no problem.