Talk:List of HDCD-encoded Compact Discs: Difference between revisions
(re: false positives) |
(→HDCD detected column with packet format: new section) |
||
(5 intermediate revisions by 2 users not shown) | |||
Line 43: | Line 43: | ||
:::C is a count of all HDCD packets; A+B are "valid" packets. Ideally A+B=C. The difference is reported in the "HDCD detected" line as "errors: N", where the packet appeared but the control data was invalid, and that is what should be used for the "Errors" column in the table. --[[User:Bp|Bp]] ([[User talk:Bp|talk]]) 22:59, 31 July 2016 (BST) | :::C is a count of all HDCD packets; A+B are "valid" packets. Ideally A+B=C. The difference is reported in the "HDCD detected" line as "errors: N", where the packet appeared but the control data was invalid, and that is what should be used for the "Errors" column in the table. --[[User:Bp|Bp]] ([[User talk:Bp|talk]]) 22:59, 31 July 2016 (BST) | ||
== King Crimson 30th Anniversary Series == | |||
According to Discogs the full series have been released several times as HDCD: https://www.discogs.com/en/label/235087-30th-Anniversary-Edition. The ones I added to the database are mainly the Japanese ones released by Pony Canyon, and while I believe that all the others should be identical to these, if someone has different pressings, it would be interesting to confirm if there is any differences between them. --[[User:Unfinished|Unfinished]] '''([[User_talk:Unfinished|talk]])''' 12:25, 1 August 2016 (CET) | |||
== Yngwie Malmsteen - Concerto Suite For Electric Guitar And Orchestra Live == | |||
@unfinished, I changed this row in the table to show that it is valid HDCD. Even if it was an accidental "Special Mode" disc, it can be decoded by the software decoder. There is probably no discernible effect, bit that is true for most/any of the pop albums, I think. Feel free to revert if you disagree. --[[User:Bp|Bp]] ([[User talk:Bp|talk]]) 17:16, 3 August 2016 (BST) | |||
:Fair enough. I actually though that software decoders could have issues dealing with it as well, but if that isn't the case, just leaving the warning about hardware decoders is fine with me. --[[User:Unfinished|Unfinished]] '''([[User_talk:Unfinished|talk]])''' 20:45, 03 August 2016 (CET) | |||
== HDCD detected column with packet format == | |||
The HDCD detected column can now show the packet format. A simple yes is still fine for this column, this is just extra information if it is available. Getting the packet format requires using ffmpeg -v verbose. If counter A or Counter B have a value, that is the format. It is possible, but unlikely that a CD could have tracks with both A and B. It would be interesting to see such a disc. It would probably be a compilation. --[[User:Bp|Bp]] ([[User talk:Bp|talk]]) 06:49, 7 August 2016 (BST) |
Latest revision as of 05:49, 7 August 2016
False Positives
Not really sure how to find them. My first though was that discs detected as HDCD but with nothing to decode (no PE nor Gain adjustment enabled) would be labeled that way. However releases like "Neil Young - Are You Passiontate" or "The Cranberries - No Need to Argue" where both features are detected in, are marked as false positives. Aren't they really used in those discs? And if so, how can they be detected? --Unfinished (talk) 16:30, 31 July 2016 (CET)
- Some albums just happen to have patterns that match the HDCD packet format. Below is the dump from the ffmpeg -af hdcd log. A normal HDCD-CD has ~30K valid HDCD packets (counter A+B) in each channel. This has 2 valid packets in channel 0, and none in channel 1, along with six sequences that appear to be packets but fail the check. This is not an HDCD, it is just coincidence that among the ~5.2M samples appeared a sequence (or two) that was a valid HDCD packet.
[274] Neil_Young/Neil_Young-Are_You_Passionate.flac ... [Parsed_hdcd_0 @ 0x14c9bc0] Process mode: process stereo channels together [Parsed_hdcd_0 @ 0x14c9bc0] Force PE: off [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control B check failed: 0x30d0 (0x30 vs 0x2f) near 24576 [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control A almost: 0x0f near 247668736 [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control B check failed: 0x18a5 (0x18 vs 0x5a) near 266510336 [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control B check failed: 0x18ec (0x18 vs 0x13) near 266543104 [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control B check failed: 0x18ee (0x18 vs 0x11) near 266543104 [Parsed_hdcd_0 @ 0x14c9bc0] hdcd error: Control B check failed: 0x16fb (0x16 vs 0x04) near 347824128 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: counter A: 0, B: 2, C: 5 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: pe: 2, tf: 1, almost_A: 0, checkfail_B: 3, unmatched_C: 0, cdt_expired: 1 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: tg 0.0: 1 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: tg -0.5: 0 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: tg -1.0: 0 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: tg -1.5: 1 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 1: counter A: 0, B: 0, C: 3 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 1: pe: 0, tf: 0, almost_A: 1, checkfail_B: 2, unmatched_C: 0, cdt_expired: -1 [Parsed_hdcd_0 @ 0x14c9bc0] Channel 1: tg 0.0: 0 [Parsed_hdcd_0 @ 0x14c9bc0] HDCD detected: no
- Thank you for the detailed answer. I wasn't aware that something like this could actually happen with certain CDs detected as HDCD, but after your explanation it makes more sense. Just to be sure I understood it correctly, the key point on your log are these lines, right?
- [Parsed_hdcd_0 @ 0x14c9bc0] Channel 0: counter A: 0, B: 2, C: 5 -> 2 packets on channel 0 (A+B, C is ignored)
- [Parsed_hdcd_0 @ 0x14c9bc0] Channel 1: counter A: 0, B: 0, C: 3 -> 0 packets on channel 1 (Again A+B)
- For comparison I analized with ffmpeg one of my rips and those lines on the log are as follow:
- [Parsed_hdcd_0 @ 0x6156e0] Channel 0: counter A: 0, B: 33926, C: 33941
- [Parsed_hdcd_0 @ 0x6156e0] Channel 1: counter A: 0, B: 33418, C: 33445
- Which seems to be in accordance with that you wrote earlier about the 30K packets per channel. On these particular rip btw, I got 42 checkfails (15 in channel 0 + 27 in channel 1), but I guess I should not worry compared to the huge amount of good ones.
- Thanks again for the clarification, I have been using both foobar2000 and Cuetools to verify my rips for a long time and while both have always agreed on the features detected, I did not check this kind of information (not even sure whether they can be configured to provide it somehow either.), so I'll stick with ffmpeg for now. --Unfinished (talk) 23:34, 31 July 2016 (CET)
- C is a count of all HDCD packets; A+B are "valid" packets. Ideally A+B=C. The difference is reported in the "HDCD detected" line as "errors: N", where the packet appeared but the control data was invalid, and that is what should be used for the "Errors" column in the table. --Bp (talk) 22:59, 31 July 2016 (BST)
King Crimson 30th Anniversary Series
According to Discogs the full series have been released several times as HDCD: https://www.discogs.com/en/label/235087-30th-Anniversary-Edition. The ones I added to the database are mainly the Japanese ones released by Pony Canyon, and while I believe that all the others should be identical to these, if someone has different pressings, it would be interesting to confirm if there is any differences between them. --Unfinished (talk) 12:25, 1 August 2016 (CET)
Yngwie Malmsteen - Concerto Suite For Electric Guitar And Orchestra Live
@unfinished, I changed this row in the table to show that it is valid HDCD. Even if it was an accidental "Special Mode" disc, it can be decoded by the software decoder. There is probably no discernible effect, bit that is true for most/any of the pop albums, I think. Feel free to revert if you disagree. --Bp (talk) 17:16, 3 August 2016 (BST)
- Fair enough. I actually though that software decoders could have issues dealing with it as well, but if that isn't the case, just leaving the warning about hardware decoders is fine with me. --Unfinished (talk) 20:45, 03 August 2016 (CET)
HDCD detected column with packet format
The HDCD detected column can now show the packet format. A simple yes is still fine for this column, this is just extra information if it is available. Getting the packet format requires using ffmpeg -v verbose. If counter A or Counter B have a value, that is the format. It is possible, but unlikely that a CD could have tracks with both A and B. It would be interesting to see such a disc. It would probably be a compilation. --Bp (talk) 06:49, 7 August 2016 (BST)