Hello,
i'm a music and audio enthusiast from Germany and love this forum as a great source of information about room acoustics.

I'm not a professional audio producer but i like to listen to music that sounds as close as possible to real voices or instruments in a recording venue.
I want to optimize my stereo system and know that the room and the placement of the speakers/seat of listener are the most important aspects of music reproduction.

The problem is that i'm not quite sure if a should try to minimize reflections from the listening room or if i should aim for a room that is more lively.

I really like the idea to listen only to the recorded instrument/voice and the recorded room. Listening only to the loudspeakers like in a non environment room or FTB rooms. I really like the concept of northward acoustics with the ideal to have an anechoic path between the loudspeaker and the listener and just give the listener some reflectons in order to not feeling uncomfortable. But is it really the correct way to listen to stereo recordings? I know that FTB-rooms are mastering/mixing rooms and you want a neutral reference. But i also like the concept for listening to music. But will it sound as close to the source as possible because you have no effects of the listening room or is the absence of reflections in the listening room a problem for stereo recordings? A paradox?
I'm talking about problems concerning the phantom sources and that reflections from the listening room can help to compensate flaws in stereo recordings. The book "Sound reproduction" from Floyd Toole describes that in great detail. Phantom sources in the center have the problem that the angle of the stereo triangle doesn't match the angle that a real source would produce (0° vs 2x 30°) and the crosstalk of stereo will produce a combfilter effect at about 2khz. Room reflections will help to fill up that dip and help to solve the problem of not having the correct HRTF because of wrong angles of the loudspeaker sound in comparison to a real source from the front. And reflections will also create a feeling of envelopment that will help to make it sound more natural. The bottom line is humans don't hear sound like microphones and reflections are necessary for stereo reproduction.

What do you pros think? Is it necessary to have some room reflections to help to compensate for the flaws of stereo recordings or will listening to the source only, without reflections from the listening room, produce sound that is as close as possible to the recorded sound?

Thanks and greetings from Germany!

You want accurate? Follow EBU Tech 3276.

Hello,

All you can do is to approach what the mastering engineer listen

So

Regarding the early reflections : for Floyd Tooles is not an issue

and

On the importance of early reflections for speech in rooms
https://asa.scitation.org/doi/abs/10.1121/1.1570439

Listen Music or Work music (mix, mastering) are two differents domains.

if you want an imitation of "the sounds as close as possible to real voices or instruments in a recording venue", the studio acoustic is not for you.


So the priority is the treatement of bass frequencies and have a decay homogeneous along the frequency range.

Compared to the multi channnel; the stereo is a poor representation of the music.

That's also the conclusion in Toole's book
But unfortunately 99.9% of all records are stereo, so my goal is optimizing the stereo reproduction. I know the basics about room acoustics and have a decent sounding room. But there are contradicting optinions about the theoretical ideal and i find that very interesting.

Despite feeling uncomfortable in an (semi)anechoic chamber, would an anechoic response speaker to listener be something to strive for even in a listening environment where the only goal is that it sounds as good as possible? Or are room reflections necessary for good stereo-listening?

What do people think that favor non environment rooms for mixing/mastering. Do you also think the rooms are ideal when you listen to music? Although reflections may reduce problems with wrong HRTF and crosstalk of loudspeakers?

For listening music, do you want a great stereo picture?

If yes the studio acoustic is not for you.

For the studio design,
The design acoustic by Northward is a way.
The design acoustic by Jens Eklund is another way.
Two different Sound aesthetics.

A test: Listen your speakers outside far from the walls, without reflection. It could help you to make your choice.

With low decay in bass Frequencies, you will need more power for the amp.

Thanks dinococcus,
i don't have to expand the soundstage with the help of reflections. Accuracy and perfect imaging is more important for me.

I am more concerned about tonality and that it could be a kind of paradox that the correct reflections in the listening room may make the sound more neutral. I'm not sure if that's correct or if the reflections do more harm than good...

In the document linked by Avare the level of reflection is - 10 Db.
Here on GS the level is - 20 Db.
Did you think - 20 Db is better than - 10 Db?
Did you think a live treatment acoustic is less neutral than a anechoic treatment acoustic?
Are you ready to use the soffit mounting solution?
Have you a listening position in nearfield or midfield?

It should not be a discussion about what i like at the moment or what i could do in my room. I've never had the opportunity to listen to music in a FTB- or NE-room so i'm asking the pros what they think I like to discuss a theoretically ideal situation and the pros and cons of reflections in a listening room.

What actually happens in a good CR is that you get "in" the music. It surrounds you. Surround (5.1 etc) has nothing over good stereo in these spaces. Low-end and sub frequencies are linear, tight, punchy and focused. All tiny variations down to the lowest notes clearly audible.

You can easily have "out of stereo field" content etc. Depth and width are incredible, yet with a strong and vivid center phantom image. You can pinpoint exactly at where an instrument is in the acoustic space of the recording. It's very much a 3D experience ... If the record is well recorded, mixed and mastered.

If it's not, the room and speakers will tell you immediately.

Yet, the space is neutral, so allows excellent translation to the outside world and you can trust every decision you take.

These types of spaces have the habit to have listeners change the hierarchy of their favorite recordings. Some they thought were great are now not that great and vice vera.

You can't reproduce this environment in a listening room at home, with tons of ER, room decay etc. These studios cost a lot of money too, and only make sense as a tool for professionals that need to work fast and to high quality standards. For them, the investment is recovered in a handful of years.

In a listening room environment at home you can find a compromise whose 'color' suits you and allows you to listen to your favorite music comfortably. It won't be reproduced exactly as intended, and quite a few elements in the recording will likely be masked. In these cases you listen as much to the speakers as you listen to the room.

But then, is it really a problem, as long as you enjoy the music?

At work, the studio is a fully loaded professional room (FTB) with ATC 300 A SL in wall, ATC 45A, 64ch analog console, Burl stereo ADC, Cranesong stereo DAC and Lynx multichannels converters, tons of fun toys to play with.

At home I have a good but simple system (ATC SCM40A + an old Benchmark DAC-1 converter for HD streaming or playback and an old Technics turntable for LPs through a basic passive monitoring interface). And zero acoustic treatment.

I enjoy both, even if differently.

@ Northward , thank you very much for your post. I really appreciate that!
I believe that you have an incredibly accurate soundstage and imaging in a FTB-room. I would love to hear that!
And obviously you interpret reflections from the listening room as a distortion of the recorded event. And i totally agree with that, at least if you consider a single loudspeaker.

But what do you think about the arguments from Floyd Toole concerning the problems of stereo-listening and that the correct reflections in a listening room can somehow make it more true to the source, because they can reduce crosstalk- and HRTF-related problems of stereo?

would love to hear thomas' thoughts on this too - my take is that it largely depends on the design principals: i wouldn't wanna miss some reflections in my lede rooms: in fact, they depend on them! but then, lede has been out of fashion for quite some time now...

Hey D,

Yes it depends on the design but to a certain extent only. For almost all designs except Moulton (from top of my head, but I might be forgetting some now) there is at least an "anechoic time gap", the "ISD gap" in LEDE/RFZ terms. And for NE and FTB, in a way you can say the ISD gap is infinite over the whole bandwidth. Or that LEDE is only temporarily semi-anechoic (floor...). Words salad.

Which means that in terms of basic overall stereo image and response while there are differences, they are not day and night... In theory.

In practice in most of the older first generation LEDE rooms and poorly implemented RFZ the ISD gap is somehow very band limited, and/or too short and/or the room builds up too much over time (room is "too long", usually around and beyond the haas trigger). And that can have a massive effect as to how the room is performing due various masking effects. In these stereo isn't great, and phantom image is somehow "selective" and weak. There is also a false sense of envelopment that can make it difficult to manage reverb, delays etc.

In a well implemented RFZ, there is no such thing. Stereo and depth, center image are sharp and clean. 3D. FTB are immune to these potential issues by design, since it's all full bandwidth ("military grade" like a friend jokes) absorption. With an ETC dropping 60dB within 100ms for an average sized room.

I don't understand Mr Toole's issue with Crosstalk and HRTF. I haven't read anything from him in a long time, so it's bit hazy now. My experience and my work and many valuable other's are showing the exact opposite. There isn't a problem with stereo in itself. I'd be out of job if ER were somehow needed to make a recording "more true" to the source. If it doesn't work, there is a problem with room design or implementation, not stereo in itself.

It's about how you decide to manage the spatial information in the recording and the mix. What reverb you use, room mic and mic setup, delays etc. How you compress and if you hear what it to does to e.g. lead vocals (too much and the vocals can start to move back in the mix). EQ.

All the info you need to make a recording "true" can be and is contained in the stereo file, without the need to impose yet another space over it, your room, for it to work.

And it can be quite spectacular

Again, when I listen to music at home while cooking, I don't care about any of that. What matters is the performance and emotion conveyed.

But in the studio, it can be an amazing experience. But it needs to translate or it's useless.

thx - also for putting things into 'historical perspective'.

maybe i should have been more clear regarding 'reflections': when i said i depend on them to some extent, i never meant to overcome 'the problems of stereo-listening and that the correct reflections in a listening room can somehow make it more true to the source' as the op mentioned: imo there is no such thing as 'correct reflections' and even less that they would make the listening experience 'more true', on the contrary!

[there was a reason why many lede designs used horn-loaded speaker systems which by design are more directional: it was to overcome some issues of poor stereo imaging! also worth noting that the massive desks used in ancient times did mess with the soundfield quite a bit, no matter how high up the mains were installed; add nearfields on the meterbridge and the nightmare became complete!

i've heard quite a few lede designs which sounded pretty 'nice' - but only until the desk and racks of outboard got move in! also worth noting that it was quite common to have the tape machine in the control room so one could do punch-ins right on the machine and not on the remote: another massive piece of gear messing with the soundfield...]

the only good thing about having some (weird) reflections was that they allowed for some sense of 'room sound' towards the 'dead end' but as much this could contribute to a somewhat better 'feeling', it often hurt the ability to hear reverb, delays etc. correctly as you mentioned.

now the funny thing is that one can get used to such a flawed design quite a bit - until one tries to go from 2.0 to 5.1 which led me to overturn the initial design principals (and rip out some of the damping in the front and add more in the back).

what's somewhat ironic is that i get to work in a venue again which was used as a rehearsal/recording space for classical music (where i assisted my mentor for ca. 15 years) and which in itself is nothing but a huge lede room! seems like i can't get away from lede, no matter what i'm doing :-)

(sorry to the op for lengthy off-topic comment - i hope all of you had a good laugh though!)

But that's not just an opinion from Floyd Toole, these problems exist. You can measure that with a "dummy head microphone" in order to have the effect of crosstalk and HRTF like humans have when they listen to sounds, and compare the response from a real source in front of it to the phantom source produced by a stereo setup in an anechoic chamber. In an anechoic chamber a recording of a single voice reproduced from a center speaker will produce a different response than the same recording being reproduced from a stereo system as a phantom source.

The question is how relevant that is from a listener perspective and if reflections that make these problems less obvious cause more harm than good by introducing other problems. And maybe a single floor reflection can mask these stereo problems so that all other reflections aren't necessary anymore...

Yes it will show a difference, but you have to look at the parameters closely and correlate to statistically relevant data from test subjects to know if it does have an effect or not. If it does, is it relevant?

The brain actually relies on synthetic data, so not on the many direct values acquired by the auditory system 1:1 but on their global interpretation at a secondary level. Depending on the conditions, it will prioritize one subset over another.

We basically have 3 main "sensors" for localization:

- ITD (Inter-aural Time Difference) ears are physically separated, so there is a timing difference within a certain limited bandwidth
- ILD (Inter-aural Level Difference) the difference in loudness and frequency distribution between the two ears
- Pinna effect (ear pinna filtering effect mostly allowing better elevation perception)

And a few secondary sensors that contribute to the synthetic data, among others:

- ITDG (Initial Time Delay Gap)
- Distance to source (i.e Loudness, bandwidth, HF content, movement etc)
- Body/Torso/Head movement as a disambiguation tool

All other things being equal, if you have a single source right in front of you in real life, ITD + ILD + Pinna + head shadow etc will have a certain set of values.

ITD=0, ILD=0, Pinna etc have your specific value at the given source elevation.

Mainly, the time and level alignment will be 'perfect' between the two ears. So the source will be perceived as coming from right in front of you.

Stereo reproduction of a single source can approximate that very well for two main factors: ITD and ILD.

Eventhough ITD will indeed have a subset of information, the head shadow + time delay of +/- 0.5ms due to the cross talk created by two sources +/-30° instead of a single 0° source.

But the general timing remains the same between left and right ear.

ITD t(0)=0 and ITD t(+0.5ms)=0.
ILD t(0)=0 and ILD t(+0.5ms)=0.

Hence there is no delta in time or level. There is no image shift.

The synthetic data gathered by the brain still tells you the sound comes from right in front of you.

It can also works well for quite a few secondary sensors (recording technique dependant):

- ITDG
- Distance to source
- respond quite well to Body/Torso/Head movement

Though depth perception can be influenced by the addition of the 0.5ms ITD. But as it is not the only factor at play, it quite easy to compensate if needed.

Adding a center speaker (LCR system) can solve some of these minor issues, but adds a number of other.

Adding room ER to that set of data does imho only add confusion and masking, not provide any further relevant information. It will only provide the brain with info about your direct environment and superimpose it over the intended presentation of the recording. That is distortion of the intended presentation of the recording.

Using a dummy head for recording content meant to be played back on a stereo speaker system makes little sense if you look at it all closely. It has artistic value for sure, but as a coloring/effect tool. If you are to playback such a recording on a set of stereo speakers, you're just adding a layer of distorsion.

I don't think we can ever create a true reproduction of a source "as if we were there". But by understanding how mics and rooms behave, the recording process, audio processing tools and recording/mixing/mastering in an environment that allows you to hear these variables we can get pretty darn close.

That's where all these high level audio engineers have such tremendous value to me. They know how to do that. It's fascinating to see them work.

That's also why it's good to always process a recording thinking many steps ahead understanding how further processing down the line will influence it all.

Just like when mixing analog, you listen through the whole chain, these days it's DAW - Analog console & FX - master ADC - master DAC -> monitoring. If you don't you almost always end up having to rework your mix.

But then 99.9% of the music is not intended to be particularly true to the original source. It's an artistic take on the sum of sources. It's meant to convey an emotion. To me, that's in the ether between the real original source and the intent.

Thanks for that detailed answer! And of course it's 100% correct concerning localization.
I'm just not sure about the tonality issues that comes with the phantom sources. That's the issue that Floyd Toole describes. He isn't talking about localization problems. Floyd Toole thinks that you need a center speaker to have a better tonality. If you don't have one maybe room reflections help to get a better tonality, which of course sounds contradictorily at first glance. But he thinks that the correct reflections above about 1khz will even out problems concerning the tonality of the phantom sources.

that's an easy test anyone can do with a surround setup: personally, i cannot confirm the finding that adding a center speaker helps 'tonality', on the contrary actually but maybe we'd need to discuss what exactly you refer to?

in my experience, there are but two advantages of using a center speaker which relate to rather to practical aspects:

- in a multi-operator situation working on a large desk, folding the mix to mono and play it back via a center speaker can mitigate the issue of not sitting in the phantom center.

- when dealing with huge channel/track counts/working on very dense surround mixes, putting some signals (other than dialogue) in the center channel can help to have less 'traffic' in the main front channels and one does not need to bother much about correlation between l/r channels either yet can still achieve a convincing 'central perspective'.

You will have a different frequency response for phantom sources in the center of the listener. Let's assume that we have the perfect microphones and loudspeakers When you record a voice in an anechoic chamber and reproduce it with ONE perfect loudspeaker in front of you, you will not be able to hear a difference between the human voice and the recorded voice. But if you play that as a phantom source by two stereo loudspeakers you will have a tonality change because of crosstalk of the loudspeakers (delay of the right speaker to the left ear and left speaker to the right ear) and different angles (HRTF, 0° vs 2x30°) compared to a sound in the center.

i admit i'm more interested in real-world scenarios such as recording in studios, on stages and in concert halls than in an anechoic chamber and hence i'm more concerned whether i perceive a signal sounding 'natural' than whether it's 'true' to the original - in fact, i don't think one can record a 'true' signal via a mic at all and even less play it back without the signal getting affected in some ways...

i find both a stereo recording (even from a coincident pair) and a stereo playback to sound more 'natural' than a mono recording and a mono playback via a center speaker - the latter i only use for pratical reasons (see my previous post).

now IF you can perceive much of a difference in 'tonality' between the center and l/r speakers AND if it bugs you, introducing some artificial reflections via delay and maybe some slight smoothing via filters is a more successful strategy: i occasionally do so when setting up/aligning large l/c/r pa systems (and provided that i'm mixing myself) albeit for slightly different reasons which do not relate much to a 'controlled' environrent of a studio...

... but in a typical small to medium sized studio, i think potential differences are way less noticeable and any additional reflections imo do nothing but blur the image/compromise the ability to precisely locate a source in the soundfield.

[sorry for my rather non-scientific comment - i weigh in having mixed some 500 albums in ca. 50 different studios and ca. 5000 live shows/broadcasts though]

I could play the devil's' advocate and say that there is no way you get a human voice to sound natural via a simple mono recording and mono playback. Even if "perfect".

Partly because the polar dispersion of voice wrt frequency is quite erratic depending on the combination of consonant and vowels, non symmetrical mouth/muscle movements etc. So a mono recording will obviously sum that, while if you stood in front of the person you would pick that up.

And partly because anechoic recordings of voice and other "natural" instruments is never a real life scenario. A natural voice, will usually come with a space attached to it. Summing the space to mono is a massive loss of information.

Vocals recorded in a dead booth are not meant to be natural sounding and get to go through massive post-production manipulations and these days, editing.

The most natural sounding dry recordings I ever made are nonetheless through a carefully placed stereo XY ribbon mic for single instruments and voice, and through XY or ORTF setups for multi instruments.

Playing them back in mono never sounded right.

The slight tonality change due to a symmetrical ITD you can manage that without problems during re-production. And it doesn't affect localization. Losing stereo field data summing to mono you can't ever recover.

All this is a compromise. Deedeeyeah's comment is what I experience too.

In the surround studios we build, it's always interesting to swap between playback of a musical recording stereo file foldback and a 5.1 source file without saying which is which.

It is quite usual to see even seasoned engineers needing to check whether or not the center channel is playing. They very often simply can't tell.

LCR also creates in theory a second layer of timing issues. Whether they are relevant or actually audible is another story.

You can do this experiment: record a vocal or instrument performance in any given space. Use 2 mics: one mono and one very well matched stereo XY. Put them all as close as possible together.

Once recorded, do this in your DAW with the stereo file:

- Copy the stereo file to a second track.
- Reverse phase and invert stereo.

Export the sum of both, this will result in a stereo file that has only the audio unique to the left and unique to right channel. All mono info is gone. Playing back that file is purely stereo info. No center. This is "Export LR".

Now go back to your DAW:

- Place the original stereo take on one track
- Place the "Export LR" file extracted earlier on another track, reverse its phase.

Export the sum of both. This gives you a file that contains only the common mono information to L&R. What's unique to L and R is gone. Playing back that file is purely mono. This is "Export Center" file.

Now in the DAW, place "Export Center" and "Export LR" (split into L and R) files on separate tracks and play with it. Notice how much stereo and location/space info you're losing with a mono mic and mono playback of a "natural recording".

(It's also a cool trick to be able to process stereo files as separate LCR files, used a lot by Mastering Engineers.)

Which one do you think is the compromise that is the most acceptable?

To me, Stereo is just a really elegant and simple way to reproduce a musical recording as faithfully as we realistically can using a fairly simple systems.

The rest is just splitting hair about something that has never been an issue and will never be an issue.

@ Northward
Thanks, i've hoped for such an answer. Actually i've never wanted to criticize stereo recordings because almost all records from my rather large collection are in stereo.
I just find it very interesting to talk about the pros and cons of reflections in the listening room and the different points of view and arguments.

Very well said!

Since this thread has now been linked and I have looked up some stuff in Floyd Toole's book in the last days, I have to say that I still see very valid arguments for the reflections in the listening room.
Maybe my example with the anechoic chamber was not good either or could be misinterpreted.

Of course Floyd Toole is not interested in reproducing mono sound sources over a center. The ambient information should also be transmitted via the surround channels.
He is interested in solving the stereo immanent problems of phantom sound sources in the center of stereo playback by the means of multichannel playback. And for stereo he thinks that the correct reflections will even out some problems of stereo playback.