Vienna Symphonic Library Forum
Forum Statistics

200,249 users have contributed to 43,194 threads and 259,062 posts.

In the past 24 hours, we have 0 new thread(s), 2 new post(s) and 73 new user(s).

  1. Home
  2. Orchestration & Composition
  3. Your Music
  4. Setting up a Standardized Wide Range Dynamic Environment
  • A
    A
    last edited
    last edited

    @Felix Bartelt said:

    So this is why it would be helpfull to know the dynamic range of e.g. the VI-16 after setting the volume faders and inserting reverb and other plugins.


    To measure every single instrument seperately, that would to be too much work, work who would not produce meaningful and helpful results for the mixer.

    The most basic and important thing for the mixer are:

    a) that the balance of all instrument track combine to a good stereo master.
    b) that at the master stereo fader no artefacts are produced during printing a mix, i.e. dB 0+ values or severe clipping.
    c) The stereo master has the required/wanted musical dynamics.

    This may sound a little dry, and as if mixing is a boring process, but everybody who mixes knows that it is an extension of composing, quasi the last arranging.

    In my case, the wanted dynamics are always “wide range,” similar to the dynamics a recording made in a concert hall or studio has. Whenever a stereo master has to be volume maximized, i.e. for tv-spots, then we do that in a second work process, this simply because we want to own a wide range dynamic master, otherwise we could process the additional maximizing plugin chain right on the stereo master fader and print the mix mastered.

    My proposed system for mixing a wide range stereo master is nothing new, but a description of what every mixing engineer work consist of since decade.

    .
  • A
    A
    last edited
    last edited

    @Felix Bartelth said:

    ...", how are the dynamic ranges of your instruments (ariving at master output) in the Sunrise_master?


    The "Sunrise" stereo master is the reference track after which the loudness table in question was made. The loudness values in the table accurately represent what the loudnesses are at the stereo master fader while printing the mix.

    I chosed this track, because the overall dynamics can be easely measured and also visually be followed on the meter and wave editor, this since the recording has a gradually raising loudness from the beginning to the end.

    .
  • A
    A
    last edited
    last edited

    @Felix Bartelth said:

    To avoid an answer like "It depens on the song…


    No, it doesn’t depend on a single song!

    I depends on an overarching dynamic concept. A dynamic concept who permitts to record and mix song after song, for years and years. And where on a future compact disc, any song, no matter if produced years ago or just yesterday, will fit between any song, and this without level adjustment or loudness modification.

    .
  • S
    S
    The original discussion thread is closed, so I'll post here and you can move it if you like.

    To me, the challenge is not total orchestra dynamics. The dynamics of the full orchestra can be compressed (for film scores) and is subject to taste. What is much more interesting to me is relative dynamics inside the orchestra. I've searched for a chart that shows dB SPL for individual instruments at several dynamic levels. For example,

    Trumpet, pp: 51
    Trumpet, mf: 75
    Trumpet, f: 93
    Oboe, f: 77
    ...etc.

    I would assume that some university grad student must have done this, but I could not find a paper after several google searches. Simply a list of instruments playing forte (perhaps muted and unmuted) would be useful. The instruments could be then be balanced in our templates. There's also the concern of relative volume versus pitch (a low forte C on trumpet will be softer than forte C an octave or two above), but this is hopefully built into the sample and outside our control in VI.

    "Use your ears" is a valid response to this question, but empirical data is always nice to start from. Perhaps VSL could provide this data on their stage if it doesn't already exist. [:)]

    Thomas J Bergersen talked about this relative balance in his column for Virtual Instruments mag. Once the relative balance is complete, then the combined dynamics can be studied.
  • A
    A
    last edited
    last edited

    @Another User said:

    "Use your ears" is a valid response to this question...


    That's what is left to do when it comes to what you call "relativ dynamics," if I understand that terminology right.

    .
  • S
    S
    That's the reason why we need to balance the instrument levels (and panning) ourselves. Even a chart that's within 10dB would be helpful.
  • A
    A
    last edited
    last edited

    @synthetic said:

    That's the reason why we need to balance the instrument levels (and panning) ourselves. Even a chart that's within 10dB would be helpful.


    +/- 10 dB SPL difference is doubling or halve the perceived loudness.

    Synthetic, and all the others… I do not see the advantage when knowing the absolute loudness and absolute dynamics of each instruments in dB SPL, and this for all register and in all relative dynamics performed from ppp to fff.

    I read quite a few time about the request to have a template for “relativ dynamics,” but I do not understand what purpose it could serve, or what exactly this template purpose could be. To me, this template would be in theory completly useless, and in practice impossible to make.

    What do I miss to understand what you mean by “relative dynamic,” and having a table with dB SPL, and the advantage to have that values in a template available for composing and mixing?

    .
  • S
    S
    last edited
    last edited

    @Another User said:

    advantage to have that values in a template available for composing and mixing?


    The advantage is that when you play a chordal pad on clarinets, french horns and cellos, the clarinets don't overwhelm the other sections in volume because that wouldn't happen in a real setting. Since sample libraries are always optimized to max volume, it's up to us to balance them to each other. A chart that said that clarinets at forte are 12dB quieter than horns at forte, for example, would allow the composer to set their clarinets to -18dB on the channel meter (allowing headroom) and horns at -6dB. When they play together, the balance should be close. For those of us who don't spend as much time with real orchestras, a balance guide would be useful. More useful to me than a chart that shows how loud the total orchestra should be.
  • A
    A
    last edited
    last edited

    @synthetic said:


    The advantage is that when you play a chordal pad on clarinets, french horns and cellos, the clarinets don't overwhelm the other sections in volume because that wouldn't happen in a real setting. Since sample libraries are always optimized to max volume, it's up to us to balance them to each other. A chart that said that clarinets at forte are 12dB quieter than horns at forte, for example, would allow the composer to set their clarinets to -18dB on the channel meter (allowing headroom) and horns at -6dB. When they play together, the balance should be close. For those of us who don't spend as much time with real orchestras, a balance guide would be useful.


    Something like that?

    Valve Horn 1th to 6th
    1st Horn distance to mic 9,0 meter
    Direction -5°
    Dynamics: ppp = -43.0 dB rms, fff = -16.5 dB rms

    6th Horn distance to mic 9,5 meter
    Direction +12°
    Dynamics: ppp = -42.0 dB rms, fff = -16.9 dB rms

    Clarinet 1th to 4th
    Distance 1st Clarinet to the mic 6,5 meter
    Direction 1st to 4th clarinet -5° to +9°
    Dynamics: ppp = -48.0 dB rms, fff = -17.1 dB rms

    Celli 9:
    Distance to microphone 2,5 meter to 6.5 meter
    Direction 50° to 65°
    Dynamics: ppp = -52.0 dB rms, fff = -17.5 dB rms

    _______________________________________________


    But what would be the parameterization for the measurement in order to make the chart?

    1) A symphony orchestra in a recording situation, who becomes your template?

    2 At which point would you measure this "relative loudness? Would the measurement be made with one stereo microphone only, or a decca tree?

    4) How wide is the stereo basis of the to be measured orchestra, or will you stay withhin the stereo basis the VSL samples have? What is the stereo basis of the VSL samples?

    5) What do you want to achiev in your mix? Is the goal to copy a recording of a real orchestra, or do we want to invent our own virtual music?

    ________________________________________________

    Even if VSL would give us this loudness information, as they are in the silent stage, the dB SPL's wouldn't be of much use to the composer and mixer; because the seating plan of a real orchestra is already a balancing; where the silent stage does not copy this situation, but rather records all instruments from an uniform distance we don't know.

    To me, working with VSL is like mixing a multitracking session with spot mics at every single instrument and group, then it is decided in the mix how loud an instrument is overall in the mix.

    I remember when the Berliner Phil's where recorded for the first time with a digital 32-tracks machine. The engineer had tremendous problems to balance the multitracks, compared to what a single stereo microphone balances almost automatically.

    .
  • F
    F
    last edited
    last edited

    @Angelo Clematide said:


    Something like that?

    Valve Horn 1th to 6th
    1st Horn distance to mic 9,0 meter
    Direction -5°
    Dynamics: ppp = -43.0 dB rms, fff = -16.5 dB rms

    6th Horn distance to mic 9,5 meter
    Direction +12°
    Dynamics: ppp = -42.0 dB rms, fff = -16.9 dB rms

    Clarinet 1th to 4th
    Distance 1st Clarinet to the mic 6,5 meter
    Direction 1st to 4th clarinet -5° to +9°
    Dynamics: ppp = -48.0 dB rms, fff = -17.1 dB rms

    Celli 9:
    Distance to microphone 2,5 meter to 6.5 meter
    Direction 50° to 65°
    Dynamics: ppp = -52.0 dB rms, fff = -17.5 dB rms

    _______________________________________________


    But what would be the parameterization for the measurement in order to make the chart?

    1) A symphony orchestra in a recording situation, who becomes your template?


    5) What do you want to achiev in your mix? Is the goal to copy a recording of a real orchestra, or do we want to invent our own virtual music?


    .



    To me something similar to that chart would help a lot. My situation is more like wanting to copy a special song than wanting to invent something new.

    So imagine I like the mix of a special song and I want to make mine sound very similar. Since I don´t have every instrument I hear in the piece, that I want to copy, playing solo, it´s hard for me, to know, how loud it is playing, cause it might sound louder, beeing supported by other instruments.
    So indeed for me it would be most important to know the relative levels, for example by someone telling me:

    At your master output (which has everything, concerning position, etc. allready calculated) a flute plays a pp at - 20 db and a ff at - 6 db. A solo violin plays..., the HO-4_marcato play at, and so on.

    I don´t see, why this should not be helpful for beginners, not having worked with real orchestras.
    So if anyone would take the time and analyze one of his example-pieces, I think, there´d be a lot of guys, beeing really grateful!
  • F
    F
    last edited
    last edited

    @Angelo Clematide said:

    IV. Dynamic Indications - Velocity to dBFS

    This PDF visualizes the MIDI velocity and the produced decibel of the four layer string patch VI-14_mV_sus_p-ff. The produced maximum peak is -7.5 dB in the ff layer, and the minimum is -36.7 dB with the pp layer.

    download link:
    http://vsl.co.at/upload/users/57/Dynamic_Indications_Velocity_to_dBFS.pdf


    .
    Sounds stupid, but I´m affraid, I still did not get everything right. What is the difference between "db RMS" and "db FS"? In my Logic master output I have a number, saying e.g. -4.1, which I consider to be the db RMS, while above there´s the peak. So when I play the VI-14_mV_sus_p-ff at maximum, where should the peak in the master output be? And if it is possible to tell this number, where is the peak for other instruments such as HO-4_sus or VA-10_mV_sus, playing at maximum in your example mix? Wouldn´t this kind of answer the question about relative levels quite easyly at least for this song?
  • A
    A

    You mention Logic, I guess you mean LogicPro on a Mac; I don’t know what the meter in LogicPro is reading.

     

    The RMS value of -4.1 is way to loud, I guess this reading displays peak. Extremly volume maximized rock music may reach -6.8 dB RMS, but that is not sounding very pleasant.
  • F
    F
    Sorry, my fault. It is the peak.
  • A
    A

    I checked Logic, it can not readout RMS. For displaying RMS you need an additional plugin.
  • A
    A

    Decibel is commonly used in acoustics to quantify sound levels relative to some 0 dB reference. The most common decibel used for music production are:

      

    RMS meter measures the voltage, which can be interpreted as loudness. In the digital domain, for example on a data stream in a digital music production software, the RMS readout displays the information about the average loudness over a defined time.

      

    dbFS is a measurement relative to digital full-scale. 0 dBFS represents the highest possible level in digital deviced.

      

    dB SPL is a measure of sound pressure level (SPL). The SPL measurement is typically used for calibrating the reference level on the monitoring in a recording studio.

     
  • A
    A
    last edited
    last edited

    @Another User said:

    And if it is possible to tell this number, where is the peak for other instruments such as HO-4_sus or VA-10_mV_sus, playing at maximum in your example mix?

    A french horn alone is also circa -12 dB peak when triggering a ff layer with 127.
  • A
    A
    last edited
    last edited

    @Felix Bartelt said:

    Wouldn´t this kind of answer the question about relative levels quite easyly at least for this song?

    No. I can't recreate or repeat the situation of how a song was mixed. A console automation of a specific song is erased no later then when a new mix is set up. We do not work with the latest consoles where everything can be recalled, we work with the console which has the best sound and this models are a little older.

    In a DAW the measurement could be excuted. But I doubt that the measured values could be of any use for a new original work. An recreating an older work it would be more efficient to put a good recording on track one, and mix the mock-up until you reach about the same mixture as the recording.

    The level of any instrument in relation to another instrument is changing very fast. A written down level value of two or more instruments is only valid for a few milliseconds. I don't see any standard there who could be put to a table. Not as a engineer nor ass a composer I think in fixed mixtures. In other word, the level of the french horns in relation to a string sections is different and unique for each situation and point in time. One time the french horn is louder then the strings, another time it is way softer then the strings; I can't see where a standard is in between that can be fixated in a table.

    There are papers which deal with relativeand absolute levels in orchestra dynamics, but not in the sense to read out and writing down the loudness values for each track on a DAW. There is an important difference between absolute and relative dynamics. Every instrument has some relative dynamic control in every register. However, some instruments in particular registers simply cannot achieve certain absolute dynamics: A group of brass playing in their high registers will never be very soft; a low flute can never be extremely loud. The best rule for a beginner is: orchestrate your dynamics instead of just writing them as textual indications. Especially at dynamic extremes, ensure that the instruments and the registers chosen are conducive to the dynamic level required.

    As a rough guide, here is a table of what the various families can achieve in absolute dynamics:

    http://www.musique.umontreal.ca/personnel/Belkin/bk.o/o3.html

    .
  • F
    F
    last edited
    last edited

    @Angelo Clematide said:

     

    In a DAW the measurement could be excuted. But I doubt that the measured values could be of any use for a new original work. An recreating an older work it would be more efficient to put a good recording on track one, and mix the mock-up until you reach about the same mixture as the recording.

    In this case I have to disagree. Specially for beginners and people like me, beeing unexperienced in orchestra music, it would be a lot of help, to have some levels to start with. I´m trying to compare my mixes to real orchestras, but since I never get every single instrument to play alone, it´s too difficult for me to tell the level for every single instrument. The difference is, that we are all using the same VSL-samples and I don´t think yours are louder than mine. Plus: A single track won´t show another volume, because it´s beeing combined with other tracks. The HO-4 track will still show a peak of e.g. - 13 db, if the strings are playing or not. So saying, it is impossible or useless to tell relative levels at a certain point of a song, seems wrong to me. All I would have to do, is to stop the song and go thru the single patches, letting them play at a maximum level, like: "At this part of the song, the VI-16_mV get a maximum peak of -12 db, while the HO-4 get..., and the FL1 gets a peak of..." Where is my mistake?
  • A
    A
    Okay, let’s assume I would write down this level values for each instrument, for example from Hans Werner Henze’s Sinfonie Nr. 1. At what points should I measure the level, and how would you apply the collected data to your music?
  • C
    C

    Hey Angelo,

        I have read the thread and have to agree with the others. Perhaps barring a release by the VI Magazine guys I could send you Thomas J's article regarding relative dynamics in order to illustrate our point. In the article he accurately describes just how important this is. His piece 'Mojo Madness' serves as further proof of just how critical this is to achieve realistic mock ups.

        Rather than debating though the artistic/scientific integrity of a template I think it stands to reason to try to gain a rough estimate. As of now I and apparently others have been unable to find even the slightest documentation in this regard. I understand how many variables are involved when recording and mixing an orchestra. I know that nothing can stand as THE template but even a rough guide is a huge step in the right direction. Once a rough guide is established one can deviate as needed while always having a safe zone or as you put it 'guideline to avoid the chaos'.

        I think the proof is in 'Mojo Madness'. One point Thomas J makes that is very interesting is that most if not all orchestral sample library's are normalized. You already brought this up by saying a PPP articulation is the same loudness level as a FFF. This is why if one were to simply measure the FF of a tutti and write from that your orchestra would still be horribly out of balance. He goes on to illustrate more complex issues such as the flute being louder in one range and softer in another. 1 flute playing and 3 flutes playing etc. He does offer Expression loudness adjustments for such occasions which was helpful. I think for now though the complexity of those issues can be set aside and one can focus purely on the maximum loudness of each instrument.

        If you knew the maximum volume of each orchestral instrument in its normal context/setting (debatable again but evaluating every variable stiffles all but the smallest of progress) than you could use a simple mod Xfade patch to determine the dynamics of the instrument once the fader is set correctly so that the samples maximum loudness matches that of the said chart we would like to create.

        As I said and I'm sure Thomas J anticipated you will ALWAYS have the science/engineering/orchestration nerds who will come forward and throw this down as idiotic or nonessential or only 'theory'. I could see the point but Thomas J has alreayd put the pie in those peoples faces. Mojo Madness is proof that this needs to exist on some level.

        If anyone has the means to make this happen or know where we can find the data please call on donations or something. This would benefit the entire orchestral mock up community greatly.

    Rob