HydrogenAudio

Hydrogenaudio Forum => Validated News => Topic started by: karl_lillevold on 2005-03-03 19:05:35

Title: Helix Fixed-point open source HE-AAC decoder
Post by: karl_lillevold on 2005-03-03 19:05:35
Yesterday, the RealNetworks Codec Team released an open source optimized High-Efficiency AAC (aacPlus) decoder to the Helix Community.

Quote
The Helix Fixed-point HE-AAC decoder provides an optimized 32-bit fixed-point implementation, which will be dual-licensed under RCSL/RPSL. No extra copyright license is required beyond the standard Helix license terms, although patent licenses are not included and must be acquired from the appropriate AAC licensors.

The Helix AAC decoder is highly optimized for ARM processors and meets or exceeds the performance of other commercial implementations. It also features one of the first widely available fixed-point implementations of SBR.

The following features are currently supported:
· MPEG-2 and MPEG-4 low complexity decoding (intensity stereo, MS, TNS, PNS)
· spectral band replication (SBR), high-quality mode
· mono, stereo, and multichannel modes
· ADTS, ADIF, and raw data block decoding
· MP4, RM, and other file formats supported via Helix client


https://datatype.helixcommunity.org/2005/aacfixptdec (https://datatype.helixcommunity.org/2005/aacfixptdec)
Title: Helix Fixed-point open source HE-AAC decoder
Post by: rjamorim on 2005-03-03 19:13:10
 

Amazing!

Big thanks to RN for contributing this.
Title: Helix Fixed-point open source HE-AAC decoder
Post by: spoon on 2005-03-04 10:22:18
Just what the community needs, thanks Karl.
Title: Helix Fixed-point open source HE-AAC decoder
Post by: guada 2 on 2005-03-04 17:04:15
Bravo. 
An evolution as one likes it 
Thank you Karl. 
Title: Helix Fixed-point open source HE-AAC decoder
Post by: yanchen on 2005-03-06 07:27:13
Great contribution for the evidence to prove the more effiective complexity design  of AAC than any other codec. From the statistical number, we can see the ARM enhanced instruction cannot bring a significant improvemnent as in speech codec. It also tell us that the comparable computation power annouced by coding technology beteewn conventional AAC decoder and HE-AAC decoder is questionable.

However, from my knowledgement base, the fastest commercial AAC ARMv2 decoder in the world can reach 18 MHZ 44.1KHz stereo. But this decoder may devote much effort between memory usage and speed.
Title: Helix Fixed-point open source HE-AAC decoder
Post by: chelgrian on 2005-03-06 20:09:04
Quote
Great contribution for the evidence to prove the more effiective complexity design  of AAC than any other codec. From the statistical number, we can see the ARM enhanced instruction cannot bring a significant improvemnent as in speech codec. It also tell us that the comparable computation power annouced by coding technology beteewn conventional AAC decoder and HE-AAC decoder is questionable.

However, from my knowledgement base, the fastest commercial AAC ARMv2 decoder in the world can reach 18 MHZ 44.1KHz stereo. But this decoder may devote much effort between memory usage and speed.
[a href="index.php?act=findpost&pid=279670"][{POST_SNAPBACK}][/a]


Hum they only tested it on a ARM9TDMI (v4T), ARM9E (V5TE), StrongARM (v4) and Xscale (V5TE). They didn't test on anything like a ARM1136 (v6) which has various media and SIMD extensions.

They also compiled it with ADS 1.2 rather than the latest ARM compiler RVCT 2.2 which produces significantly faster code.

The fastest ARM ACC codec was handcrafted in assembler by some guys down the corridor from me but we no longer license it commercially.
Title: Helix Fixed-point open source HE-AAC decoder
Post by: Bonzi on 2005-03-07 01:31:04
Hi Karl!  This is really awesome contribution to the open source community.  But is the RCSL/RPSL license compatible with GPL?  I ask because the it would be really nice to use this code in the iPodLinux project.
EDIT: Nevermind, according to this (https://helixcommunity.org/content/complicense) it is .
Title: Helix Fixed-point open source HE-AAC decoder
Post by: The DJ on 2005-03-07 17:02:29
Quote
Hi Karl!  This is really awesome contribution to the open source community.  But is the RCSL/RPSL license compatible with GPL?  I ask because the it would be really nice to use this code in the iPodLinux project.
EDIT: Nevermind, according to this (https://helixcommunity.org/content/complicense) it is .
[a href="index.php?act=findpost&pid=279923"][{POST_SNAPBACK}][/a]


It also says this:
1 Note: because this license contains certain reciprocal licensing terms that purport to extend to independently developed code, You may be prohibited under the terms of these otherwise compatible licenses from using code licensed under their terms with Covered Code because Covered Code may only be licensed under the RealNetworks Public Source License. Any attempt to apply non RPSL license terms, including without limitation the GPL, to Covered Code is expressly forbidden. You are responsible for ensuring that Your use of Compatible Source Licensed code does not violate either the RPSL or the Compatible Source License.


so basically RPSL is compatible with GPL, just not viceversa (according to the strict explenation of the GPL license).
Title: Helix Fixed-point open source HE-AAC decoder
Post by: HotTrance_COM on 2005-03-12 08:01:18
Great!  Now we just need an open source HE-AAC ENCODER

~ J
Title: Helix Fixed-point open source HE-AAC decoder
Post by: Garf on 2005-03-12 11:37:37
Split Nero on Linux discussion to:

http://www.hydrogenaudio.org/forums/index....showtopic=32285 (http://www.hydrogenaudio.org/forums/index.php?showtopic=32285)