- #21 [LaLaLam], 23-04-10 19:17#20, 無聊師兄,MQA 的確跟 HDCD 理念接近,HDCD係90年代中的產品,佢係用左 digital audio 16 bit 的最低 LSB 的1bit 去 control 同辨認是否 HDCD, 通常這個 LSB在普通 CD係不停變化的 floor noise,HDCD就 hard code左做1, 於是佢地開發的那粒 pm100/pm200就會認到呢個不變的LSB就會著左 HDCD的指示燈,進入pm chip的解碼狀態,所以事實上 play HDCD係只有 15 bit, 但佢地粒 pm chip本身係 DSP,所以可以做多好多嘢,1.開動壓縮解壓程式(如下圖示),2.用特別設計的 dithering filter,於是佢地話兩者加埋等如20bit dac, dynamic range>115dB, 不過一般認為這只是一個 trick而已,這間公司最後俾 microsoft 買左,Windows player聞說還有 hdcd plug in, 同樣 foobar也有一個類似的解碼 plug in.
- #22 [無聊會員], 23-04-11 21:42#21
LaLaLam兄,
謝謝解釋.
Windows Media player 同 foobar2000 有 HDCD plug-in, 我知道但未用過. 很久很久前, 我有部 Audio Alchemy DDE v. 3.0 係有解HDCD功能, 不過未用過, 因為我冇HDCD碟, 又係很久很久前我有用 Masterlink 9600 自己整24bit CD玩,依家就絕大部分係file, up & down conversion 就一直都係部古董 dCS 972.
另外想請教 MQA file 用FLAC封装,又冇乜嘢原因? - #23 [LaLaLam], 23-04-11 22:23>.... MQA file 用FLAC封装,又冇乜嘢原因?
如果是 end user,咁 MQA 用 flac 還是wav其實沒所謂,兩者都可以正常認到,音質分別也不大,不過 flac format本身有 Metadata tags, 可以包含歌名,封套 artwork 等和用甚麼 encoder (e.g. MQA or HDCD),所以如果用 music editor的話,咁 flac 的 tag 指向 MQA 本身既 metadata就變成必須,而且不會弄壞 MQA裏面隱藏了的data。 -
- #24 [wklie], 23-04-12 00:45#22: A correctly implemented MQA decoder will work on all lossless formats including WAV, FLAC, AIFF, ALAC and MQA CD. Some implementations fail without proper tag, but that would be a design flaw.
最後修改時間: 2023-04-12 00:46:52 - #25 [LaLaLam], 23-04-12 08:10>... but that would be a design flaw.
Actually, this may not be a design flaw. Indeed, the MQA file structure is quite different from a normal LPCM file. Ref. Pic. below, a 24 bit LPCM is linear 24 bit count from LSB, However, the MQA coded 24 bit has to break into 8 bit + 16 bit with two different LSB and be processed separately. FLAC tag can be used as the function of the red arrow as shown. - #26 [wklie], 23-04-12 09:18#25: Although a few MQA partners adopted this wrong design in their streamers, it is clearly wrong because it will fail several scenarios.
The simplest examples are this:
1. Use a CD transport, connect SPDIF to MQA CD player - the FLAC tag is not passed via SPDIF. With this wrong design, the MQA CD decoding cannot be invoked. A correct MQA CD player will handle it (e.g. Esoteric K-01XD).
2. Use a computer to play a free download of 24-bit FLAC MQA to a USB MQA DAC. The computer software will decode the FLAC to PCM. There is no FLAC tag when sending the MQA PCM to the USB DAC.
最後修改時間: 2023-04-12 09:19:17 - #27 [DY2036], 23-04-12 09:36好少人談論就係如果因為隻碟花而導致訊號損失,噉響呢種情況之下MQA CD或者MQA CD Player會點樣做,對輸出音效會唔會有可聆聽嘅失真。
- #28 [LaLaLam], 23-04-12 10:26這也是一個問題,CD Red book訂立的規格是八十年代的產物,從來沒考慮會有pseudo random noise coding (PRN) 例如MQA的存在,而 CD 本身的Cross-interleaved Reed–Solomon coding CIRC 當然不會顧及 PRN,sub-coding 是有機會互相影響的,兩者其實是需要有學術機構行出來做一個整體規劃才能完美化,而家 MQA 進入了 chapter 11, 相信會有大哥出來打救的,而家大哥中出PRN coding最多最成功的在東方,不過 MQA細眉細眼,應該不感興趣吧。
- #29 [kx14cp1], 23-04-12 11:31所以一隻無花的碟,一部讀取正確數據的cd唱盤,對普通cd或mqa cd嚟講一樣重要。步步高讀碟皇幾花都讀到,但有幾多信息係原本定估算出嚟就唔知。
- #30 [jimmyihv], 23-04-12 23:19MQA is in debt too much and the future is not good. I think it is not worth to save it.
Besides Tidal will have Hi Res music and MQA will not beat Hi Res music at all.
最後修改時間: 2023-04-12 23:21:56 - #31 [無聊會員], 23-04-13 16:23
[#23]
LaLaLam兄, 謝謝解釋。
[#24]
wklie兄, 謝謝關於decoder說明。
講過decoder, 又睇下一個有趣嘅encoder故事. 呢位GoldenSound係Tidal嘅contributor, uploaded咗啲MQA encoded files上Tidal做testing, 點知比Tidal鏟files同account, 去Tidal同MQA投訴, response就哈哈哈. 呢位GoldenSound去咗Audio Science Review, disclose呢單嘢. 據呢位GoldenSound做嘅test同analysis, 出咗呢啲意見, 有興趣可以睇下, 對digital audio有D認識就會容易理解D, 因為有D似學術論文.
聲明: 本人從來未講過MQA是好是壞, 本人同LaLaLam兄嘅討論集中係接近消失嘅HDCD.
https://www.audiosciencereview.com/forum/index.php?threads/mqa-deep-dive-i-published-music-on-tidal-to-test-mqa.22549/
0:00 - Intro
1:15 - Summary of topics
1:47 - Pre-video clarifications
3:44 - What is MQA?
5:41 - How I got test files encoded in MQA
6:57 - MQA 44.1khz vs Native 44.1khz
8:14 - MQA HiRes vs Native 88.2khz
11:11 - Reduced Dynamic Range
11:47 - Does unfolding work? (Many MQA releases are just upsampled 44.1khz)
12:41 - No Lossless on tidal if track is MQA
13:47 - MQA "Authentication" means nothing
14:30 - MQA Upsampling filter is leaky
15:33 - Filter attenuation analysis
16:54 - HiRes unfolding analysis
18:53 - Full decode analysis
19:54 - Conclusion
21:19 - My Opinion, and testing cables
25:46 - Why are manufacturers supporting MQA?
27:13 - Best way to play MQA files
27:44 - ifi GTO filter is MQA
28:56 - MQA's response
37:02 - Credits/Thank you
0:00 - Intro
0:45 - 1: Removal of my tracks
1:17 - 2: MQA's claims of "Losslessness"
4:06 - 3: Provenance and the Blue Light
5:42 - 4: Real world music vs test signals
6:54 - 5: More Real world music vs test signals
8:20 - 6: MQA Encoder errors
9:42 - 7: MQA's feedback
10:23 - MQA Editing wikipedia page
12:25 - 8: MQA Adding distortion and aliasing
14:00 - Are MQA's screenshots fake?
16:07 - 9: MQA sound quality and deblurring
16:37 - 10: MQA encoder is only for musicc
17:03 - 11: Video Title
17:15 - 12: General comment
18:30 - Closing thoughts - #32 [LaLaLam], 23-04-14 11:18#31,無聊師兄分享的投訴MQA事件十分有趣,值得細味和觀察,有關MQA這類形的投訴非常多,甚至有部份用家話某種類形的音樂是不適合用 MQA去編碼解碼的,例如用電子合成樂器,快速掃頻的效果器等,要測試和認証這方面的問題是否重覆性是十分困難的,是需要一個有權威性的學術機構去作第三方測試才能知道問題所在,尤其現代的 audio streaming係經過網絡去完成,而且普通家庭用的”non-deterministic networking”; 數碼訊息是經過無數的 server/router , time/frequency domain 上也存在大量非確定性,MQA的前途的確不太樂觀,反觀飛記在 80年代初已在 CD 規格訂立了類似HDCD/MQA 的編解碼(pre-emphasis / de-emphasis) 存在,到現在部份 CD ripper亦需要去處理呢個很少用的flag. 我手上其實也有頗多這類古老CD, 佢地都係用 14 bit AD/DA,所以需要用此壓縮/解壓的方法去擴展動態。
- #33 [無聊會員], 23-04-15 21:55#32,
LaLaLam師兄, 關於現代嘅 data streaming, 請教Lam師兄, streaming video & audio 之類嘅data, 我理解係行Internet Protocol 下面果啲 Streaming Protocol, streaming 係用封包 packet, 呢個封包化對啲 video & audio file嘅還原有乜影響? 謝謝。 - #34 [LaLaLam], 23-04-16 20:51簡單而言,Audio Video Streaming 係用TCP/IP 內的 UDP 包裝成比較細的 packet,目的是減低 network latency 網絡時延產生的影響,而且可以利用 streamer內建的 jitter buffer去還原訊息,如果用家是用自己內聯網 stream自己的nas 問題不會太大,但如果是外聯網 stream 距離非常遙遠的 server, 咁呢個 latency 的影響就會非常嚴重,所以專業用的 protocol 本身係有 time stamp去控制每個 packet都放在準確位置,而網絡亦需要用上所謂 TSN (time sensitive networking).
- #35 [evga56], 23-04-17 08:01#34
audiophile 聽歌又唔係realtime application,只要個jitter buffer夠大就得,有latency 對音質有乜影響? - #36 [PhantomGTR], 23-04-17 08:43係同一件 hardware 內,Streamer 用了 low latency kernel, or even real time kernel 會好聲好多。
Latency 大,呢聲會死好多,暗唖好多,音場、動態好多嘢都會差好多,直接影聲音嘅音樂感、真實感同現場感。
最後修改時間: 2023-04-17 08:47:25 - #37 [VinylBear], 23-04-17 14:18Well... 最近幾天在youtube見到有鬼佬講, TIDAL都有可能轉用lossless FLAC. 真係唔明白 現今世界hi speed internet已經咁普及, 點解TIDAL仲要用lossy compression方法去做streaming, DAC又要有MQA... 又麻煩又低能.
- #38 [無聊會員], 23-04-19 17:51#34
LaLaLam師兄, 謝謝解釋.
另外想請教Lam師兄, audio steaming 個 packet size 雖然好細, 但可能有啲 packets 會遊地球幾個圈先至去到房企個 router 入 streamer, 會有 packet loss, 呢啲 loss 要去到幾多%先會影響個file, 同係家用房企環境下, besides buffet, 有冇乜嘢方法同工具去改善呢個 loss 問題 ? 謝謝. - #39 [LaLaLam], 23-04-19 21:16Audio streaming 其實算是real time application 所以不能用有 error而 resent 的 TCP 而需要用有 error 即 lost 的 UDP, 所以packet loss 的確會做成失真,劣化s/n ratio 等大問題,所以現代的 streaming service 都會在發射和接收端加入forward error correction code (FEC). 簡化圖如下,即是如果一個 packet loss, 咁佢只影響一大組 data的其中一小部份,之後從redundant parity byte restore 整組 data, 即有改錯功能,有D似CD 機沿用致今的 cross-interleaved Reed–Solomon code (CIRC),但就算有 FEC 改錯功能,如果 packet loss太嚴重的話,其實都會影響SNR, 下圖的 code gain 就成為packet loss vs 音質的指標。
- #40 [evga56], 23-04-21 16:54#39 Coding gain嘅SNR同轉analog 之後嘅audio SNR有咩關聯? 實際bit error rate高,package/frame/MSB lost已經係爆音
- 下一頁 (2 of 3)
- 返回 ...