In Reply to: Reading comprehension an issue? posted by Charles Hansen on December 19, 2005 at 21:26:14:
Well, what can I say, I did a search for all posts with HDCD authored by myself and immediately found my previous post - see belowThere was a technical paper on HDCD available on the web that described the HDCD encoding and decoding process - unfortunately this no longer seems to be available but it was previously on http://www.hdcd.com/partners/proaudio/articles.html
Sorry - unfortunately I did not save a copy of this paper. Perhaps you can try and get a copy by contacting Microsoft directly.
For reference, this is what I posted in April, which is a summary of my conclusions from reading that paper:
my understanding is that an HDCD encoded CD differs from a "normal" CD in at least three different ways:
1. a proprietary dithering algorithm that supposedly gives a "perceived" extra 3 bits of resolution
2. plus a way of encoding an extra bit, or 6dB of dynamics
3. plus a way of encoding dynamic switching between multiple filters on playback (which is why HDCD players are equiped with a special digital filter chip)if you play an HDCD on a player with no HDCD support, you still get the benefit of (1) but not the next two. You also get the music automatically "compressed" by a factor of 2:1 due to the "missing" 6dB of dynamics (that's why HDCD often sound louder than normal CDs when played back without HDCD decoding)
i think the Microsoft HDCD decodes (2), not sure if it properly decodes (3). Presumably (3) is done by requantizing into 24-bits, but an even better process is to upsample 2x or 4x at the same time (which the Microsoft HDCD decoder unfortunately does not do).
*** It's really swell that "TI and Cirrus Logic have DSPs that support HDCD decoding, along with higher sample rates for normal PCM" now. I can assure you that they did not in 1998 when we were developing the Ayre D-1 DVD player. ***
But we are living in 2005 (soon to be 2006), not 1998, so what's your point?
By the way, last I checked, nearly all the major manufacturers support HDCD on their flagship processors/receivers and/or players: Denon, Pioneer, Rotel, Harman Kardon, Panasonic, Sherwood, ...
Very few of these implementations are based on the Pacific Microsonics chip, so if you think the only way to implement HDCD today is to use Pacific Microsonics, you should look further.
Essentially HDCD these days is implemented as software on a DSP and no longer relies on hardware. Think of it as a proprietary upsampler based on hints (the HDCD command set) supplied by the encoder.
Given that it's just software, there is no "cost" or engineering constraint to implementing it, apart from paying the license fee. Your objections to HDCD are out of date, you should really catch up and live in 2006 rather than 1998.
This post is made possible by the generous support of people like you and our sponsors:
Follow Ups
- Maybe you could do with a crash course on how to use search as well? - Christine Tham 23:11:37 12/19/05 (43)
- You want an apology? - Charles Hansen 22:17:21 12/20/05 (38)
- Perhaps it is you who has reading comprehension issues? - Christine Tham 00:59:31 12/22/05 (18)
- This is just silly - Charles Hansen 09:08:41 12/22/05 (17)
- Who is being silly? - Christine Tham 11:49:06 12/22/05 (16)
- That would be you - Charles Hansen 20:27:20 12/22/05 (15)
- Re: That would be you - Christine Tham 15:23:38 12/23/05 (14)
- Re: That would be you - Charles Hansen 20:14:53 12/23/05 (13)
- Re: That would be you - Christine Tham 00:01:07 12/24/05 (12)
- Figure 3: Digital rip of an HDCD without Peak Extend - Christine Tham 00:09:39 12/24/05 (9)
- Figure 4: Output of WMP (no Peak Extend) - Christine Tham 00:11:08 12/24/05 (8)
- By the way, here are some stats on the non decoded vs decoded outputs - Christine Tham 00:28:38 12/24/05 (7)
- Thanks for posting these graphs - Charles Hansen 10:51:35 12/24/05 (6)
- Re: Thanks for posting these graphs - Christine Tham 12:12:00 12/24/05 (5)
- This is just silly - Charles Hansen 14:46:22 12/24/05 (4)
- HDCD is defined by the patent, not the implementation - Christine Tham 22:38:00 12/24/05 (3)
- Oh, that's right... - Charles Hansen 13:06:23 12/25/05 (2)
- And your point is? - Christine Tham 15:31:48 01/03/06 (1)
- Yep, yep, and yep. - Charles Hansen 19:43:56 01/03/06 (0)
- Figure 1: Digital rip of an HDCD with Peak Extend - Christine Tham 00:04:18 12/24/05 (1)
- Figure 2: WMP decoded output of the track - Christine Tham 00:06:28 12/24/05 (0)
- Charles you do owe Christine an apology, because you are dead wrong. Christine has it 100% correct. - Teresa 06:51:43 12/21/05 (18)
- Hmmm... - Charles Hansen 09:34:25 12/21/05 (17)
- Charles I have found some links to help you understand HDCD better: - Teresa 19:38:30 12/21/05 (3)
- Re: Charles I have found some links to help you understand HDCD better: - Charles Hansen 20:07:24 12/21/05 (2)
- Thanks for the links Charles, it will take a while to read all of these. - Teresa 20:46:09 12/21/05 (1)
- You're welcome - Charles Hansen 22:00:32 12/21/05 (0)
- Still waiting for that apology ... - Christine Tham 16:52:05 12/21/05 (12)
- You are still confused - Charles Hansen 19:52:19 12/21/05 (11)
- Re: You are still confused - Christine Tham 00:01:40 12/22/05 (10)
- Taking a page from W - Charles Hansen 09:11:58 12/22/05 (9)
- Re: Taking a page from W - Christine Tham 11:54:11 12/22/05 (8)
- I guess we agree on one thing - Charles Hansen 20:03:13 12/22/05 (7)
- Re: I guess we agree on one thing - Christine Tham 15:25:13 12/23/05 (6)
- Re: I guess we agree on one thing - Charles Hansen 20:16:22 12/23/05 (5)
- Re: I guess we agree on one thing - Christine Tham 00:38:47 12/24/05 (4)
- Once again you are wrong - Charles Hansen 11:00:41 12/24/05 (3)
- I don't think you know the meaning of the word "wrong", since you are not applying it correctly - Christine Tham 12:25:41 12/24/05 (2)
- You have gone beyond obstinence... - Charles Hansen 18:49:44 12/24/05 (1)
- Sorry, but you haven't really given any evidence - Christine Tham 22:47:18 12/24/05 (0)
- I did find the HDCD patent (#5,479,168) - Christine Tham 23:30:02 12/19/05 (3)
- Christine -- give it a rest. You were interesting once. nt. - Martin419 02:58:46 12/21/05 (2)
- I'm sure Christine will as soon as Charles apologizes. She is 100% correct BTW. (nt) - Teresa 07:00:12 12/21/05 (1)
- Thanks, Teresa (nt) - Christine Tham 17:03:58 12/21/05 (0)