AC3 to 6-Waves.32Bit for audio editor

[NoX1911]

I'm trying to decode an AC3 file to 6 waves.32bit, change something in center channel and re-encode to ac3.

1) I tried BeSweet1.4 (produces one file only. Seems not to work at all.) and BeSweet 1.5b31 (-6chfloat) with BeLight. The produced wave files cannot be loaded with soundforge or acid. Audition can import but data is not correct (double speed, distortions,...) though it is interpreted correctly as 32bit/float.

2) Headac3he converted the ac3 file to one DDWav that gets recognized correctly by audition as 6ch, wav, 32bit. But it is not possible to load it in acid (for ac3 encoding) and i don't want to produce useless huge intermediate files.

Which audio editors can handle BeSweet's 32Bit waves or is there any other compatible AC3 decoder to produce 32bit 6 waves files?

[Skelsgard]

BeSweet has a high quality resampler associated tool (SSRC) but has some issues handling 32bit files.
I'm trying to decode an AC3 file to 6 waves.32bit, change something in center channel and re-encode to ac3.
AC3 files are 16bits. Why would u want or need to resample to 32bits if the final destination is a 16 bits as your source is?.
i don't want to produce useless huge intermediate files.
Sometimes u must if u want to ensure quality. Besides, 32bits files occupy 2x the diskspace 16bits files do.
I would recomend u to just decode a Six Waves with Belight or BeSweet GUI without resampling and load them into Soundforge or Audition, then resample to 32bits within this programs (they do high quality resampling), change whatever u have to change and resample back to 16bits, since the final AC3 is gonna have to output 16bits no matter what bitdepth the source has.

[NoX1911]

Are you sure AC3 is 16Bit? I thought its like MP3, neither 16bit or 32bit (non-pcm?). But to get best/nearest results you have to choose 32bits.

Nevertheless.. seems like BeSweet's 32bit output is for internal processing only and not for external programs so i have to use 16bit.

[foxyshadis]

That is technically true, but 16 bits is already around or often below the noise threshold of the codec. The inaccuracies of the dct-fitting are usually much higher than the sampling precision, anything higher than that is random static (although it's way below the threshold of hearing). Very few lossy codecs are designed to usefully output higher resolution sound, and it'd need a very high bitrate anyway.

If you're going to be further processing (reverb, eq, etc) 24/32-bit does make sense though.