HydrogenAudio

 Hi, everyone

This a traditional poll.
There is also 2018 Format poll [LOSSLESS] (https://hydrogenaud.io/index.php/topic,115721.msg954780/topicseen.html)


Results from previous years
(https://s18.postimg.org/u82z6vma1/lossy_2017.png)

Why can i choose only one option? I listen regularly mp3 and aac radio shows, podcasts etc, but never encode with them.

Wait: are there any USAC encoder which I can use freely?

1. WavPack Hybrid is missing
2. Just one option to choose? I regularly use three lossy codecs.

Wait: are there any USAC encoder which I can use freely?

No, not to my knowledge. There is a thread about USAC and it is still only available to radio broadcasters and similar.

Why can i choose only one option?

2. Just one option to choose? I regularly use three lossy codecs.

I assume this is for consistency with all the past years. I, too, use multiple codecs (AAC, MP3 and a little Opus). I would like more of a points/weighted option like I have 5 points to give based on how much I use them. AAC would get 3, MP3 1, and Opus 1. But again, that would break the ability to compare with past years.

Other than .flac (lossless),  I am really stuck using .opus.  I have a specific example where .mp3 given its design/parameter choices about masking, actually removes a subtle, but important part of some music that I do testing with.   Opus does better (still removes some of the subtle sound), but isn't perfect either even at 442k.  However, given the fact that Opus does do better for me even at the same rate as maximum legal mp3, I have to use opus (but also try to use a lossless format first.)   (One specific example is a well DolbyA decoded version of 'A Kind of A Hush' from the Carpenters, where the vocal is actually echoed with a very close echo.   That echo virtually disappears using .mp3, is simply just a little weaker/a little distorted than uncompressed using .opus, and perfectly clear (of course) using .flac.  BTW, the echo is almost impossible to hear (if not totally impossible to hear) unless the material is 100% pristine.  The example (actually made for someone else who I am working with) will be on the following site very temporarily (probably another week):

 https://spaces.hightail.com/space/YCvYjaBRFX

I haven't posted .mp3 or .opus versions, but if someone is interested, they can play with the source material which the actual version which would be "01 -- There's A Kind of Hush-decoded" and the one where mp3 has some troubles.

John

If at 442k you still notice a flaw, isn't it due to the codec or has it something to do with the resampling? You can try to resample first using different methods en see if that eliminates the problem. After that you can encode to Opus.

New engine of the forum doesn't support multiple choice in poll.
Ask admins if it's possible to change this.

It's only possible if someone wants to code it. The developers of the forum script apparently aren't interested, or they'd have done it by now.

I use a MP3 at V0, opus at 128kbps, Musepack at 170k with the nmt switch for to up the bitrate on more demading stuff. Since i like MPC more i made that my main choice and what i would use if i got a new android phone.

(VBR) MP3V0 if need be. It's eminently compatible.

MP3 (some formats never die)
ACC (iTunes purchases)
Dolby Digital (TV broadcasts, DVDs)
DTS (some CDs, DVDs)
OGG Vorbis (Spotify)
Opus (some Youtube Videos)

I prefer Vorbis (Aotuv) as main codec for regular music.

- New lossy music files: Vorbis ~256kbps
- Existing lossy files from the last few decades: A mish-mash of formats and bit-rates. I don't transcode if not necessary
- New digitized vinyl and music cassettes: A homemade ADPCM codec 384kbps (slightly better than that what I used before for these)
- Existing digitized vinyl/MC: IMA-ADPCM 4Bit (works better than most modern ones for these in my opinion, because I don't want to  denoise them and I want to add a 60's style pseudo-stereo in most cases so I get the sound I'm used to as I grew up. Both is poison for psychoacoustic codecs)
- For mobile use: Vorbis, 32KHz,  64-80kbps
- Audio track of homemade video: MP3 and AAC (each 256kbps)
- Speech: Opus ~32kbps

1. WavPack Hybrid is missing
2. Just one option to choose? I regularly use three lossy codecs.

1. You're right. I haven't included it because previous polls indicated very low share (~0-1%).  Now I see there is some discussion about Wavpack lossy in 2017 poll. We should include it next time.  As for now it's late, sorry.
2. Already answered above.

If at 442k you still notice a flaw, isn't it due to the codec or has it something to do with the resampling? You can try to resample first using different methods en see if that eliminates the problem. After that you can encode to Opus.

It isn't the sample rate -- if I use any sample rate at 44.1k or above, the echo is still there.  It appears to be that the algorithm for removing material (in the mp3 and less-so opus) that would normally be masked is removing the small 'echo.'  (Not really an echo, but a delayed replication of the signal.)

Normally, one probably would not hear the difference -- unless being incredibly critical.  Opus is truly better in maintaining the content -- but slightly modified/distorted (not distortion in the intermod/harmonic sense, but a less clear rendition of the echo.)  Frankly, I wish mp3 at 256k or higher would be as good, and I wish opus was perfect at 442k, but there has to be a loss of information when the amount of data is reduced by a factor of 10.  (That idea is NOT the reason why I might THINK that there is a loss of detail -- it REALLY is there.)

Now, regarding resampling -- I always use sox with the rate -v command for changing sample rate, so that is pretty darned good.  If I used a more primitive resampling mechanism, that might also lose some info -- but I doubt it.  The kind of loss that I hear would be attributable to masking choices in the software.  I could probably make a tweak somewhere to make the masking less aggressive (even though it is already okay 99.9% of the time, even for mp3.)  I simply need perfection for some of my testing, and lossless and the terrible amount of diskspace is needed :-(.

If I was just listening to the music, I probably would not have heard the difference.  But, I have my 'measurement ears' on, which is a different way of hearing.

I don't really use any lossy codec on a regular basis. I only use FLAC but I have few MP3 albums for music I yet have to find in lossless (hard to find Italian CDs in USA) and when I convert for my wife's iTunes I use AAC.

I'm using AAC @ 160kbps for listening music on my phone but I'm ready to jump to Opus @ 96kbps once Android or Samsung supports Opus on their .opus container instead of mkv. That was a poor choice.

I'm also going to join the ranks of the group complaining about the poll letting you choose only one option, because that is not what happens in (our) reality: specially amongst us long-term HA 'format browsers'.

I've been told polls can be configured to let people vote N times, and that lets you choose multiple options, and that's the official word. I don't even know if they had the sense to restrict it to one vote per option, or let people vote whatever one choice they want multiple times. Not good.

Of course, we could have continued to use the same unsupported forum script for another decade, it definitely supported the correct voting options for polls. Nobody was openly attacking IP.Board 2 forums anyway, so not likely anyone would have exploited a hole that probably didn't even exist.

Plus, playing whack-a-mole with the spammers constantly, including through the invention of a custom moderation script to do things the forum didn't support natively, kept us all on our toes.

The poll shall have more choices because I use on regular basis different formats for different devices.

The poll shall have more choices because I use on regular basis different formats for different devices.

I guess he's asking for the one you use "the most" on a "regular" basis...

spotify had recently became available in my locale, so it's all Ogg Vorbis for me from now on.

I'm having a hard time considering streaming services a viable choice for this thread. It just doesn't match the whole idea of this poll with localy stored media that you made the choice for it's format. For example, in iTunes or Amazon, atleast you do have files localy stored and you can re-encode them to something else if you have a preference to another format. To make it even more complicated, if I said that I was using Deezer, it offers 2 options of codecs for streaming, mp3 and flac. What if I said that i was listening my music from Youtube, would I vote for opus, aac or ogg? Also what if a streaming service offered somekind of proprietary codec? The whole idea behind streaming services is that you are not given any kind of choice in codecs, just qualities. Also, if Spotify didn't get pressured to share the fact that they use OGG, you wouldn't know that it's using and they want as much as possible for you to have no idea what they are listening to. So the problem here is mostly that you are given no choice and you do not possess any file to begin with. The final and bigger problem is that it also clashes with what most of the people in the poll had in mind when they voted for themselfs. We need in some way to separate Streaming Services from this topic.

I'm having a hard time considering streaming services a viable choice for this thread. It just doesn't match the whole idea of this poll with localy stored media that you made the choice for it's format. For example, in iTunes or Amazon, atleast you do have files localy stored and you can re-encode them to something else if you have a preference to another format. To make it even more complicated, if I said that I was using Deezer, it offers 2 options of codecs for streaming, mp3 and flac. What if I said that i was listening my music from Youtube, would I vote for opus, aac or ogg? Also what if a streaming service offered somekind of proprietary codec? The whole idea behind streaming services is that you are not given any kind of choice in codecs, just qualities. Also, if Spotify didn't get pressured to share the fact that they use OGG, you wouldn't know that it's using and they want as much as possible for you to have no idea what they are listening to. So the problem here is mostly that you are given no choice and you do not possess any file to begin with. The final and bigger problem is that it also clashes with what most of the people in the poll had in mind when they voted for themselfs. We need in some way to separate Streaming Services from this topic.

A lot of people don't have locally stored music at all anymore. They rely 100% on streaming services. Not including them, would be actually what's skewing the statistics, because I'd argue this is where most of the use for a codec comes from these days: The ability for it to be streamed.

Sure the choice is not one of using a codec to encode music from a lossless source, but it's still very much a market decision on the user's side. I can see why some users would choose one service over the other, because of the codec/quality.

When it comes to streaming services that let you select the codec, you'd choose your preferred codec here. Keep in mind, this is the poll for the "lossy" codecs, so selecting FLAC even though Deezer offers it, would be incorrect anyhow. But that is also a non-sequitur.

I use youtube for listening to music all the time. And it's super easy to see what codec you're using, especially when using mpv + youtube-dl. While mpv and youtube-dl display it right there in the console, even when you're using the youtube website you can right-click on the video and check what kind of media is playing. (right click -> "stats for nerds").

When it comes to services like Spotify: assuming they wouldn't want to tell you the codec: At the end of the day, that audio stream needs to be decoded on your device. Incases where you're using a browser on a desktop or laptop, it'd be super easy to figure out what sort of codec they'd be using. If they opt for something proprietary, and need in-browser decoding in JavaScript: well, they're in for a ride... When it comes to apps on a phone, etc.: You could still intercept the bitstream and inspect it. Essentially, for a media provider to be secretive about their codec, is kindof a fool's errand here. Those kinds of secrets, would be more than likely figured out sooner or later. That tangent aside, you'd simply choose "other" from the list, in case that is truly the most used codec that you'd use here.

But I also get what you're getting at, and I think the misunderstanding comes from the term "use". "Using" a codec, is very much using a streaming service in my book. However "using" a codec to encode from a lossless source, is also - well - "using" that codec. So Perhaps one might want to compartmentalize that further, but that Is something on the end of the person running the poll. I believe @IgorC intention is to get an overview sort of result, and I believe that includes streaming services, DAB, etc, etc.

The poll includes the option "USAC / xHE-AAC", which as of now, I know of no freely available encoders. I actually know of only one streaming service available on the internet, that streams in xHE-AAC. I wish I could listen to that stream on my computer - for testing and suchlike - but there's simply no decoder available to my knowledge. All other streams are short-wave or medium-wave DAB radio (which require horrendously priced radio receivers, and I know of no one personally who owns one). As far as I can see it, the only way to "use" USAC right now, is to either listen to it with a special DAB radio, or being a radio station yourself, having access to the encoder software and/or appliance hardware.

With streaming services, the user, arguably, doesn't use a format, unless it's a conscious choice.
It's used when you encode something in it for your own needs.

Hmm… it could probably make sense to define the meaning of "using a format" in the opening post.

in short, I voted for... Opus.

Was AAC (AAC-LC) until very recently as I had a old phone laying around (Huawei y336-a1) which with Foobar2000 mobile it works great for Opus which I generally use in the 64-96kbps range since that's one of the major points of using Opus is strong sound at lower bit rates as once you hit the 128kbps+ range your probably better off using Apple AAC.

while I realize MP3 is a safe format, because of wide hardware support, it's outdated mostly because Opus/AAC are superior especially at lower bit rates. basically AAC makes MP3 mostly outdated because it's more efficient (quality sound at lower bit rates (96-128kbps)) and still a widely supported format which is the only thing going against Opus is lack of wide hardware support like MP3/AAC. but assuming your only going to use your lossy files on hardware you know Opus supports, then Opus is best simply because it's the best lossy format straight up when you put aside hardware support.

Wavpack hybrid .  Archive = External usb . PC & android phone = wv lossy only @ 352k -x4.   You encode ONCE and wv + wvc files are created so no transcoding or multi archives whatsoever. Also its lean even on vintage hardware like android 2.3 ARM. Works very well with foobar mobile and poweramp.