CD will not sound better than lossless rips, and data retrieval from HDD is orders of magnitude quicker and more reliable. CD is a data carrier nothing more, better more reliable ones exist. The people you talk about probably just understand the theory and facts, and know why this isn’t the case.
6 Likes
yes also the the project transport. I have my eye on that one.
Do you really think people who disagree never listened to a CD in their life?
3 Likes
Not entirely true. The bits are the same and in the same order until you factor in jitter. Jitter is the difference in arrival of a bit over successive bits (bits 1 & 2 arrive with spacing of 3 and bit 2 & 3 arrive at spacing of 4 you have jitter of 1). If jitter is not controlled it can cause a D/A to treat a 0 as a 1 as the voltage swing happens at the wrong time. A digital path’s ability to reduce jitter can and will influence the quality of the analog output. This is also one of the reasons why different inputs on the same DAC can sound vastly different.
Not really useful in streaming audio though. If you’re waiting for a webpage to load you wait for the lost packets to be resent and you just think the site is slow. With anything that must be consumed “in real time” you have no time to wait. That wait is variable based upon how the receiver notifies the sender which segments are lost and then waiting for the sender to go send them again. In my experience this doesn’t happen fast enough and so, by the time the retansmit occurs, the time to play those few milliseconds of music are past us and they get thrown away. This serves only to create additional network load. Why use TCP then? It’s a good, cheap, data metric to understand how closely the far end is processing packets to their send time. If the far end is having issues the sender, in this case Roon, will know quickly and can notify the user there are issues with that endpoint.
The difference is that Tidal, Qobuz, and others front-load the music stream as fast as possible. If packets are lost towards the end of a song and you’re still listening to the start of that song then you have time for retansmit to occur. RAAT doesn’t work this way though. RAAT is streaming very close to the portion you’re hearing.
It depends on the source. Tidal and Qobuz at their highest resolutions do not do this. It’s a fixed rate using a FLAC container.
Most all encoding CODECs support a variable bitrate option including AAC and MP3. You can also reencode from fixed to VBR. I don’t know any provider streaming with VBR but (some early internet radio stations did though), as you mention, some will jump rates based on connection quality or setting (Spotify lets you set bit rate differently for cell vs. wifi in their app). Tidal and Qobuz do not do this though. They simply will fail the stream. But, and I found this funny, Tidal will periodically reset the streaming quality on mobile app upgrade. I can only assume this saves them considerable bandwidth as most people can’t hear a difference.
Be careful here. My digital playback system is much better than my CD system. Could I go buy a better CD transport and then say my CDs sound better than streaming? Sure, but why would I do that when I live in a streaming world? If I want better streaming I’ll spend the money on a better digital transport and playback path. A “good” CD transport is approaching the best streamers out there in price and with the CD transport you’re limited to redbook. I’d much rather own a top of the line digital transport (streamer) than buy a top of the line cd transport in 2021.
6 Likes
Okay, I understand that variable rate streaming doesn’t exist. I use Qobuz. I am comparing to digital downloads (on a disc attached to a Roon Nucleus) not to playing CDs. The chain is identical from the network switch to the speakers. The difference is the router and the satellite modem, with data capped at 25mb/s. It does take some time to load a track for streaming and if you play an album there can be a failure along the way. But the sound is also less good than the digital download. It is okay for pop music. It is not okay for classical.
In this case it isn’t the DAC. The path to the DAC is the same with a digital download or streaming–Nucleus, streamer, DAC.
Path jitter is only relevant for synchronous data (S/PDIF, I2S); Ethernet or USB are asynchronous and buffered. However, the conversion from asynchronous to synchronous within the DAC may be jittery, for sure. That’s a main reason for differences between different DAC inputs.
Broadly speaking, people often blame upstream components for the flaws in their DACs, in part because they are loath to accept that their pricey brandname DAC has poorly designed or implemented digital receiver circuitry.
7 Likes
Indeed they do but I don’t know of any significant platforms that support VBR music streaming, while it’s pretty ubiquitous in video streaming.
My findings are that this 600 CD transport via audioquest diamond digital spdif cable sounds vastly superior spinning the same disc as streaming that same thing through Tidal. I have a Roon Nucleus + hooked to my DCS Rossini DAC/clock combo for streaming. My first post was mostly as a curiosity if running from nucleus to wifi hub to DAC via ethernet was the reason my streaming quality is less than spinning the disc in a relatively inexpensive transport via spdif into the DAC. I would have thought someone was crazy also saying that before I tried it. I was looking for ideas on whether or not my setup was lacking something that was causing me to hear such a difference. Mostly I was looking for any advise on if the link where the ethernet runs through the wifi ethernet switch might be the culprit or if running via USB from the Nucleus + sounds much better. Ethernet is the way DCS says to hook it up. I didn’t realize everyone was going to get their panties in such a ruffle about this. Over here at my system it sounds way effing better running the CD than streaming.
1 Like
I think this might be the ticket to try. Go USB to the DCS DAC for streaming and keep the mesh router/switch out of the signal path. I think there might be a few conveniences given up running USB instead of ethernet but might be worth a try. Was looking for some of this kind advise rather than mud slinging.
I hate to be that guy, but it’s not the first time that a really expensive DAC proves to have mediocre circuitry on some of its inputs, especially those that need more complex asynchronous processing like USB or Ethernet. TL; DR Even fancy DAC makers may cut corners.
4 Likes
That would be called Adaptive Bit Rate if you mean vary according to network conditions, not Variable Bit Rate I believe.
Just to confuse things, you will come across ABR with two different meanings - with AAC and MP3 it means Average Bit Rate where the bit rate varies around an average in order to maintain quality and there is Variable bit rate where the bit rate varies according to music data complexity (so lowers quality where it think it can get away with it).
In video streaming (netflix etc) there another ABR which is Adaptive Bit Rate that varies according to network conditions. Then there is Multi-bit rate which I believe is just multiple different bit rate streams.
I am not aware of a commonly used adaptive bit rate mechanism for music streaming.
3 Likes
I don’t have the resources to figure this out but I suspect latency to the data center where the track is coming from makes a difference. Mostly, my tracks start as quickly via streaming as local. I have very short RTT to US west coast data centers though. For example I’m ~40 ms right now to Google us-west-1 and less to AWS similar region.
This 100% 
It’s also why people should really experiment with the different inputs on their DAC.
Our hearing is much better than our eyes. Most people don’t notice the tricks used for the rate change of the video. You’d absolutely hear the rate change if this occurred on a piece of music. We use Adaptive BR for voice calls all the time though… we’ve got some good tricks to reduce bandwidth usage without significant changes to the quality of a call. However, those CODECs don’t work for music or test signals. They are specifically tuned for people talking.
Not at all. Your next experiment is to find a SPDIF output from a streamer. It could simply be the network input on the DCS isn’t great. The USB test is free but most USB implementations are not great.
I… hmmm… if I came across that way I apologize. You’ve got a new target for sound quality. I believe that level of sound quality is achievable via streaming. However, you didn’t ask for recommendations so this whole thread is unsolicited advice in that direction and I have also contributed to that.
3 Likes
You are indeed right, my mistake.
This…
I sold my quality $3000 CD player over a decade ago because ripped music sounded better. Now with MQA and dsd, it’s even better. The weakest link is the USB connection if that’s what you are using.
I don’t believe anyone has mentioned that volume normalisation applied by the streaming services could be a reason for differences heard between a CD transport/player and streamed music. Check out this site
Loudness Penalty: Analyzer
The site does not state as much but it will analyse FLAC files
Good point, but I guess it gets applied to rips/downloads as well if leveling is on.
And what people use themselves and then forget.
Its 1s and 0s, if its the same data is sent from a cd, ethernet or wireless the difference is not in the data delivered that is different but in how the DAC interprets those 1s and 0s to express the sound. Thats it. I like the analogy that if if there was loss or degradation of the data based on the delivery hardware banks would be out of business. CDs may sound better than streaming in only that the masters are different or the dac is interpreting ing the sound differently,. If no change in resolution happens then they are the same. The idea that routers or electronic interference degrade those 1 and 0s is suspect at best, the data is the data. These analog opinions out on Digital material is just eh …
1 Like
Late to the party again…Hey now you are talkin’ about something I’m a fan of vinyl. Have you ever tried ripping vinyl and processing the files to 384/32 bit float? Try playing the hard file (no roon dsd) in native hi-res hi-bit-depth on your DAC and see what you think. IMHO hi-res hi-bit-depth digital vinyl is the next best thing to dropping the needle.
1 Like
Just my musings…
In my rig.
1/ Vinyl first and foremost, period.
2/ Metal cassette tape recorded on Nakamichi zx7 from vinyl.
3/ hires Qobuz stream, usually upsampled via Roon to DSD 256.
4/ Regular Qobuz stream native resolution.
5/ Ripped CD played back via Roon.
6/ Straight CD playback.
Now to qualify though.
Well over 10k vinyl gear.
Well over 4k analog cassette gear.
Well over 5k streaming/DAC.
Old oppo bdp103 universal player.
Would a 2k CD player up the ante?
Maybe.
Not going to find out though…
1 Like