How to compare high-res audio streaming to CD playback?
One of the most crucial questions for an audiophile is usually: is it better?
Unfortunately, I think there is no correct answer, but it will be the subject of another article: I prefer to make decisions based on long term experience against short term comparisons.
Audiophiles like to compare generally, but making meaningful comparisons is not easy in the age of lossless streaming.
Since music streaming providers announced their new lossless or high-resolution streaming services, social media has a recurring question: will it be better than CD? Comparing CD sound to a sound of a streaming service is very difficult to make in any meaningful way, as one should compare two usually completely different audio subsystems.
From a production point of view, post-production of CDs and files for streaming is the same until the mastering. After that, however, the masters can be slightly different as most streaming providers have loudness recommendations for their normalisation process, which are not always honoured by the labels.
The master could be saved in FLAC with the same sampling rate and bit rate and then uploaded to the streaming provider in most cases or it could be converted to 44k/16 bit and then uploaded to the streaming provider if they require lossless but not high-resolution.
The CD manufacturing process starts from here based on the 44k/16bit format. However, as the master is usually higher than 44,1kHz/16-bit audio files, 48k/24 bit or higher, there will be a conversion from the master on CD in all cases. CD manufacturing involves an electro-mechanical process (music CDs are pressed) for making it (pits printed or moulded in case of CD-Rs) and an optical-electrical process and error correction on the playback side before the D/A conversion. Playing a CD means following the spiral of pits and lands. Playing CDs back requires all sorts of processes, like error correction, optical reading, reading head controlling through servo, etc. In addition, a well-designed power supply is a must to minimise distortions on the D/A conversion part of the playback process. Thus, playing CDs always involves motors, servos, spinning discs.
Technically, a high-res stream from Qobuz will be “closer” to the master than any CD. However, Tidal and MQA are debatable. I will discuss it in another article. I like MQA, which is misunderstood as a ”codec”, while it is an end-to-end process by most of those who do not like it.
If anyone likes CD better at home than streaming, it is possibly caused by the local system or its optimisation, limitation, not by the streaming technology as such.
Streaming companies are content managers. They have their own or hire cloud storage (from Amazon, Google, for example) and operate the streaming servers, subscription and revenue sharing system behind it. Most of them do not do anything with the files uploaded by the labels, right holders, or so-called music distributors. Even normalisation is done at the end of the chain by their apps, not in the cloud.
So if you hear a difference between using a CD and listening to a stream, I think there are few potential reasons, but not the streaming and cloud storage technology itself:
- The masters are not the same. Even 0,5 dB difference here and there can make a difference. There is a chance that, in this case, the overall loudness of the track will be different too (so-called LU(FS) value). None of us knows if the masters are different, just a little bit.
- Gain structure is different in the different paths. Therefore, these comparisons are not completed with the same checked and calibrated loudness.
- Personal bias
- Different playback chains.
- Tidal (Qobuz) are not in exclusive mode or using normalisation. If you need to make a comparison, do it in Roon. It can decrease the number of variables a bit.
- Streaming services have a bit different loudness recommendations than a CD is made. You can see them in the chart 3 below. They need it because of their normalization process. There is quite a bit of a difference between the average CD and let us say Apple Music.
- If you hear a sound quality difference, it could be caused by so-called adaptive streaming, but according to my knowledge, none of the streaming services uses it for music and audio. Netflix does this. You can see that sometimes the first few seconds of the movie are much worst than the rest of the movie because they start to play immediately in lower resolution while the buffering goes on.
So I think there is no technical reason why streaming should sound worst than a CD if you do not use normalization and use your PC in exclusive mode. Tidal normalizes audio to an integrated -14 LUFS but can be set to quieter -18 LUFS. So it is important when one uses Tidal, to switch normalization off and set the DAC usage to exclusive. My experience is that most of those who felt the CD sounded “better”, were using Tidal in a trial mode, and not with hifi subscription, so they could listen to only the compressed version of the tracks, not in MQA quality.However, PC-based streaming can be strongly influenced by the audio subsystem of your PC, how the memory content is fed to the audio output devices. Tidal and Qobuz continuously fill a buffer and plays from the buffer. No spinning disks or servo-controlled heads are involved.
Optimising a streaming-based system can be tricky.
An optimised CD-based system can rely on the knowledge of the manufacturer of your CD player/DAC.
One should have to be more involved to optimise and get the most of a streaming system as it involves some networking skills.
*Link to the original chart: https://www.masteringthemix.com/blogs/learn/76296773-mastering-audio-for-soundcloud-itunes-spotify-and-youtube
