Yes, and probably without advance notice that would give you to time archive or download
Streaming vs Casting
Shoutcast and IceCast are classic streaming services. There is a directory of audio streams organized by genre. Anyone who selects a stream gets the same content as everyone on the same stream. There is no way to select albums or songs. Shoutcast has a license to block players having free software modules so the VLC media player has dropped Shoutcast. Fortunately, IceCast provides free streams like Shoutcast once did. However, the directory is a mess. There are a few gems and many lumps of coal. The problem is to find the good ones. VLC shows a big list of IceCast streams. The search feature is good. For example, enter "HearMe" and see a list of Jazz, Blues, Ambient, etc. Right-click on any item to see the metadata.
I am puzzled by this streaming. Why would anyone bother to do it. Who are they trying to serve. How is it costed.
Spotify and other well known streaming services are not actually streaming... they are file serving. Nothing wrong with that. I suppose they did not know what to call it.
Shoutcast and IceCast are classic streaming services. There is a directory of audio streams organized by genre. Anyone who selects a stream gets the same content as everyone on the same stream. There is no way to select albums or songs. Shoutcast has a license to block players having free software modules so the VLC media player has dropped Shoutcast. Fortunately, IceCast provides free streams like Shoutcast once did. However, the directory is a mess. There are a few gems and many lumps of coal. The problem is to find the good ones. VLC shows a big list of IceCast streams. The search feature is good. For example, enter "HearMe" and see a list of Jazz, Blues, Ambient, etc. Right-click on any item to see the metadata.
I am puzzled by this streaming. Why would anyone bother to do it. Who are they trying to serve. How is it costed.
Spotify and other well known streaming services are not actually streaming... they are file serving. Nothing wrong with that. I suppose they did not know what to call it.
Absolutely! I do wonder how many die hard streaming enthusiasts have actually heard a decent CD or vinyl system and whether they would continue chasing nirvana if they did.
In recent versions of librespot (the open-source Spotify library that's backing so many non-official frontends) we first "promote" a 16-bit signal to greater bit depth (well, floating point actually) by scaling it. Then we do software volume control, normalisation, dynamic compression / limiting on it (all optional) before converting to whatever output bit depth you choose -- 16/24/32 bit integer or 32/64 bit float. Advantage being lower quantisation error of any aforementioned DSP pipeline.
Hi @roderickvd. Is there any way to circumvent this "promotion". Is there any way to compare a play with and without promotion so the value of the process can be experienced.
Indeed - but until recently I did not know it. A couple of years ago I had one of those Audio Nirvana moments that you never forget.
We were installing an L'Acoustic line array PA in a school "Cafetorium." You know, one of those spaces that is a gymnasium, dining hall and has a stage. Anyone familiar with these knows how horrid the acoustics are. This one was typical, and it was not sounding good. I suggested that the speakers needed toe-in, that firing at the back wall wasn't helping anything. So I got up in the lift again, and gave them some toe-in. The system designer did some tweaks on the signal processor. The crew chief plugged his phone into the mixing and played "Money for Nothing." Holy Crap! It was stunning. Much bigger than life. Super clean, detailed, the tom fills went across left to right, each the size of a 55 gallon oil drum, but perfectly separated and placed. Sting's high, light voice floated off high and to one side, Knopfler's strange sounding guitar riff came in from the other side, big and solid and powerful. All so real, so powerful, solid and bigger than life - but also with amazing placement in space. Just stunning. But it was late, the install was done and he cut off the music. I never got to hear it again.
All that from an MP3 (I assume) out of an Android phone. Sometimes the speakers can reveal incredible scenes even in lowly formats.
I used to think so too, until I got Tidal. (I am one of those Roon users. LOL) The new music constantly discovered makes it well worth my money. And sound quality is important to me. (I have Hifi Plus at $19.99/mo)
An example of something I would never have found on my own. There is so much music on Tidal I could explore it for the rest of my life and never hear it all. Stuff I really like I am somewhat likely to purchase and add to my own local file library.
I can certainly see that. And that is the one reason I still consider streaming services. I used to discover new music by listening to the radio. That seems to be a thing of the past, especially as shows/stations that feature classic rock are fading into the ether.
Tom
I had Nirvana, then I upgraded the bearing in the LP12 and got a P2.
Now, I have Nirvana... and I'm getting a P3.
It's a hobby. Our main home system long ago ceased to "sound like a stereo" and it "sounds like music"... but, it just gets better and better... Actually, we got five full time systems...
But, I got to tell you... I no longer have a "CD player". With Tidal HiFi, Android and good DACs, I get really good streaming quality too. And with the latest move away from MQA into FLAC, Tidal is getting better.... but, note, it's a subscription... you got to pay for the quality.
That would be a great experiment but unfortunately not really. It would require a fair bit of work to change all the code to go from normalized float to integer.
Best you could do is try librespot 0.4 or better even the development version versus 0.1. The latter does not normalize to float, however lacks dynamic limiting / compression.
My view is that it would probably take golden ears to hear it, however that there’s no reason to do “the right thing”. When we did some cursory testing on the performance impact from doing this, we found it to be negligible on CPU load and memory usage even on a lowly Raspberry Pi Zero.
Anecdotal evidence may be found with e.g. the SRC419x series of resamplers. While their input maxes out at 24 bits, internally it has 28 bit processing. Same for most DAWs that process in 32 bits. Quantization error is cumulative in a pipeline, so the more processing, the greater the gains (or actually: the lower the losses).
Final note that comes to mind regards decoders. While I’m no expert on them, I do know that for many (not all) there is no given relation or symmetry between input and output. Meaning if you input e.g. 16 bit integer to the encoder, for the decoder it may output x bit integers or floating point all the same. Stored in its compressed form, the bit relationship is lost. (Be aware to separate this audio bit depth from kbps bitrate.)
Final-final note: these 2 cents are about bit depth advantages for DSP, not about fidelity of recordings.
Last edited:
Classic rock sounds rather dull when recorded with Lo-Fi gear. Remastering does not help. I find I need to crank up the volume because of low dynamic range. The solution is to hire a cover band who can perform the original perfectly so that high-fi masters can be created anew with modern gear. To name a few, Pink Floyd and Alan Parsons Project need to be redone.
Originally FM radio sounded as good as CDs. Here in Toronto GTA there are so many stations jammed in the FM dial that it sounds trashy. It is ok for podcasts or news, not music.
Here are some FYI links.
Spotify recently revealed that white noise and ambient podcasts rack up 3 million hours of listens a day.
- https://www.wired.com/story/universal-music-deezer-war-on-streaming-noise/
Hobbyists stream pseudo white noise or mindless nature recordings noise over mp3 from Spotify. This is technically absurd since one can properly generate white noise with a Zener device or a low-cost IC (Here I oversimplify). These streams are stealing resources needed for artists and spread disinformation about White or Pink Noise PsychoAcoustics.
- https://www.wired.com/story/universal-music-deezer-war-on-streaming-noise/
Hobbyists stream pseudo white noise or mindless nature recordings noise over mp3 from Spotify. This is technically absurd since one can properly generate white noise with a Zener device or a low-cost IC (Here I oversimplify). These streams are stealing resources needed for artists and spread disinformation about White or Pink Noise PsychoAcoustics.
When subjected to well-controlled double-blind A/B testing, very (very) few people can.
Proper 320kbps Ogg Vorbis encodes are excellent. The whole "compressed audio sucks" argument is based on crappy 128kbps MP3 encodes from back in the days of Napster.
Do I still listen to 44.1/16, 96/24, etc. FLACs from my personal library? Of course! Do I want to pay more to stream lossless? No way. Streams are for BGM and music discovery. Heck, even 128kbps AAC on SomaFM works fine for these purposes. Find music you love, and then acquire it and put it in your library.
Like Tom, I can't imagine being a slave to some big corporation for access to my music.
Last edited: