Hi,

Parallel perceptual encoding can't be done in that way (splitting in time
domain, and then merging the result) because psychoacoustic model and other
tools need to know data from previous frames as well as look-ahead data.

The idea of parallelization could be done like following:

1. Splitting the algorithm in two parts - PSYCH (filterbank+psychoacoustic)
and LOOP (quantizer + huffman coder)

2. PSYCH module would use few frames look-ahead (i.e. - process few frames
before LOOP starts quantization and coding)

3. One CPU (or DSP) would do PSYCH and other CPU would do LOOP part of the
algorithm

Of course, this needs a lot of tweaks and synchronization code,

Best,

*************************************************
Ivan Dimkovic, Technical Manager

PsyTEL Research
Multimedia Coding Solutions
Belgrade Yugoslavia

phone:  +381 63 264 334
phone:  +381 64 11 40 600
fax:       +381 11 32 25 275

email:  [EMAIL PROTECTED]
www:  http://www.psytel-research.co.yu
*************************************************
This e-mail may contain confidential information which is legally
privileged. The information is solely for the use of the addressee named
above. If you are not the intended recipient, any disclosure, copying,
distribution or other use of the contents of this information is strictly
prohibited. If you have received this e-mail in error, please notify us by
return e-mail and delete this message. Thank you.

----- Original Message -----
From: "Stefan Ohlsson" <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Monday, March 25, 2002 9:33 AM
Subject: Re: [MP3 ENCODER] lame parallel encoding


> Hi,
>
> On Sat, 23 Mar 2002, Christian Schoenebeck wrote:
> >I read someone succeeded in modifying the blade encoder to be capable for
> >parallel encoding. AFAIK was the result not byte for byte same as decoded
in
> >the serial way but sounded normal (no clicks or something).
> >
> >Has somebody already succeeded with doing that with the lame encoder?
> >
> I haven't tried, but as I read this I thought of the --nogap option in
LAME.
> Would it be possible to chop up a large WAV file in pieces of approriate
> length, encode them at the same time and then merge them to one large MP3?
> Not quite. LAME needs to know about the previous WAV file.
>
> But what if I split my WAV into, say, 4 WAV files and encoded them in two
> batches: 1,2,3 to 1.1, 2.1, 3.1 and 2,3,4 to 2.2, 3.2, 4.2.
> If the size of these WAVs is selected with care, not much extra work needs
to
> be done on the files that will be doubly encoded (1,4 big, 2,4 small).
> Would it be possible to merge 1.1, 2.1, 3.2, 4.2 to one big MP3?
>
> /Stefan
> --
> [ Stefan Ohlsson ] � http://www.cixit.se/ � [ 020-4702019 /
+46-709-561314 ]
>
> _______________________________________________
> mp3encoder mailing list
> [EMAIL PROTECTED]
> http://minnie.tuhs.org/mailman/listinfo/mp3encoder

_______________________________________________
mp3encoder mailing list
[EMAIL PROTECTED]
http://minnie.tuhs.org/mailman/listinfo/mp3encoder

Reply via email to