setting the gain levels (1 Viewer)

[Telúrico]

Hello!

I make audio recording of birds and publish them in eBird. I have a few doubts. Hope you can help!

I use a Zoom F6 and a M66 mic or a telinga parabolic and i have no knowledge of sound techniques... I record at 32 float and between -17/-20 db in the external knob. The result is a very low sound! But i have been forced to do it because the eBird requires a -3db normalization on edition and setting higher levels in the external knob produce a lot of noise. I´m thinking right? Am i wrong? For some raison low sound recordings before normalization don´t seems right

Thank you for any explanation!

Francisco

 

[Jon.Bryant]

As you are recording in 32bit float the recording levels are not that important (the Zoom F3 32bit float recorder, doesn’t even have a gain control).

Basically 32bit float means there is lots of headroom (a big difference between the loudest and quietest thing you can record). You can also record above 0dBfs (decibels full scale) - in 16bit and 24bit the loudest thing you can record is 0dBfs (represented by a value of zero). With 32bit float you can have negative numbers - basically representing sounds louder than a value of zero.

The 32bit file should be normalised and then best converted to 16 or 24 bit using DAW software (I think Audacity will suffice, which is free).

I don’t think you should really see much difference in noise regardless of how you set the gain. If you set the gain high, then you may get some pre-amp noise, but then you are probably not multiplying this by much when you normalise to -3dB. If you record a very quiet signal, then you may not hear too much pre-amp noise, but whatever noise there is, will be multiplied when you normalise the recording. It should basically be swings and roundabouts with 32bit float.

It is a bit different with 16 and 24 bit recordings (particularly the former). The more bits the larger the number that can be stored and with digital audio, larger numbers mean quieter sounds. With all digital formats the quietest sound you can record is called the noise floor. If you set the levels far too low, then parts of the signal may fall below the noise floor - you would need a bigger number than can be held by the format, in order to represent parts or all of the signal. If this occurs, when you normalise the recording you will be amplifying the noise floor, but cannot recover the signal lost below this level. With 32bit float the noise floor is much lower that even 24bit, so you would need to set the levels incredibly wrong to ruin into this problem.

The man issue with noise is generally when you have a low signal to noise ratio - say a distant and quiet bird in a noisy environment or recorded with noisy equipment. There is no real way to cheat a poor signal to noise ratio - when you apply gain or later normalise the signal, the ratio of noise to signal will persist. You can try to reduce noise with DAW software, but the better software is not free. Better to position the mic closer to the bird, to get the best signal to noise ratio possible.

I don’t think -3dB is a must, but advisory. Basically listeners don’t want to have to turn up the volume for one recording and then be deafened on the next. This is why the platform suggests normalising to -3dB. If you have a very quiet recording of the main subject this may not be possible, as to normalise to this level will result in a very noisy recording - ambient noise will have too much gain applied in the normalisation process to sound acceptable. I would suggest normalising to a level less than -3dB and checking what sounds tolerable. Perhaps very quiet recordings (which cannot be easily normalised without creating lots of noise) are not really worth posting online, unless they are of something really unusual.

 

[Jon.Bryant]

The result is a very low sound! But i have been forced to do it because the eBird requires a -3db

Reading this again I am a bit confused. -3dBfs would mean that the signal should peak on the F6 ballistic meter just below the 0 mark (well half way between the 0 and 6 mark below). I am surprised if you could get near to this with a very low gain level, after all -3dBfs is pretty loud. I would have thought that you would need to have a high recording level (which may create audible noise) to get the meter to peak near -3dBfs. In my thoughts, chasing -3dBfs is more likely to mean you are aware of noise during monitoring, rather than vice versa.

As said in my last post, normalisation is done in post. If your recording is quieter or louder than -3dBfs (and has a good signal to noise ratio) don’t worry, just normalise if afterwards. The EBird requirement should not force your selection of recording level - particularly when you are using 32bit float. The beauty of 32bit float is that it is very forgiving and you can fix things later.

 

[Jon.Bryant]

Just looked at the F6 manual, as I don't know this recorder. As this is a quite complex multi-track recorder, the knobs have different purposes in different modes. When recording in 32bit float mode, the default is for the knobs to work as gain control for each channel prior to pressing record, but during recording the knob are faders for the left right mix - tracks 1,3,5 are mixed into a left channel and 2,4,6 into a right channel, and the faders control the loudness of each channel in the mix. In this default setting, you cannot change the actual channel gain once a recording has started. You can change this default behaviour so that the knobs always act as gain control (and the LR mix faders are bypassed). I this this is Menu>Input>Track Knob>Rec Level (rather than Reference Level).

I presume that you are not using the recorder as a field mixer, as you say you are using it with a ME66 (mono shotgun) or Telinga Parabola (either mono or stereo depending on model). I think you can therefore ignore the LR mix and work with the track 1 (and 2) files, which should be the unmixed signals from your mic (assuming the mic is connected to those channels). You can set up inputs 1 and 2 as a stereo pair, if using a Telinga Stereo mic.

Before considering the LR mix, I would make sure that all unused channels are disabled - I presume that if you are accidentally applying lots of gain to an empty channel, you could still be adding preamp noise to the mix.

 

[Telúrico]

As you are recording in 32bit float the recording levels are not that important (the Zoom F3 32bit float recorder, doesn’t even have a gain control).

Basically 32bit float means there is lots of headroom (a big difference between the loudest and quietest thing you can record). You can also record above 0dBfs (decibels full scale) - in 16bit and 24bit the loudest thing you can record is 0dBfs (represented by a value of zero). With 32bit float you can have negative numbers - basically representing sounds louder than a value of zero.

The 32bit file should be normalised and then best converted to 16 or 24 bit using DAW software (I think Audacity will suffice, which is free).

I don’t think you should really see much difference in noise regardless of how you set the gain. If you set the gain high, then you may get some pre-amp noise, but then you are probably not multiplying this by much when you normalise to -3dB. If you record a very quiet signal, then you may not hear too much pre-amp noise, but whatever noise there is, will be multiplied when you normalise the recording. It should basically be swings and roundabouts with 32bit float.

It is a bit different with 16 and 24 bit recordings (particularly the former). The more bits the larger the number that can be stored and with digital audio, larger numbers mean quieter sounds. With all digital formats the quietest sound you can record is called the noise floor. If you set the levels far too low, then parts of the signal may fall below the noise floor - you would need a bigger number than can be held by the format, in order to represent parts or all of the signal. If this occurs, when you normalise the recording you will be amplifying the noise floor, but cannot recover the signal lost below this level. With 32bit float the noise floor is much lower that even 24bit, so you would need to set the levels incredibly wrong to run into this problem.

The man issue with noise is generally when you have a low signal to noise ratio - say a distant and quiet bird in a noisy environment or recorded with noisy equipment. There is no real way to cheat a poor signal to noise ratio - when you apply gain or later normalise the signal, the ratio of noise to signal will persist. You can try to reduce noise with DAW software, but the better software is not free. Better to position the mic closer to the bird, to get the best signal to noise ratio possible.

I don’t this -3dB is a must, but advisory. Basically listeners don’t want to have to turn up the volume for one recording and then be deafened on the next. This is why the platform suggests normalising to -3dB. If you have a very quiet recording this may not be possible, as to normalise to this level will result in a very noisy recording (even ambient noise will have too much gain applied in the normalisation process to sound acceptable). I would suggest normalising to a level less than -3dB and checking what sounds tolerable. Perhaps very quiet recordings (which cannot be easily normalised without creating lots of noise) are not really worth posting online, unless they are of something really unusual.

Just looked at the F6 manual, as I don't know this recorder. As this is a quite complex multi-track recorder, the knobs have different purposes in different modes. When recording in 32bit float mode, the default is for the knobs to work as gain control for each channel prior to pressing record, but during recording the knob are faders for the left right mix - tracks 1,3,5 are mixed into a left channel and 2,4,6 into a right channel, and the faders control the loudness of each channel in the mix. In this default setting, you cannot change the actual channel gain once a recording has started. You can change this default behaviour so that the knobs always act as gain control (and the LR mix faders are bypassed). I this this is Menu>Input>Track Knob>Rec Level (rather than Reference Level).

I presume that you are not using the recorder as a field mixer, as you say you are using it with a ME66 (mono shotgun) or Telinga Parabola (either mono or stereo depending on model). I think you can therefore ignore the LR mix and work with the track 1 (and 2) files, which should be the unmixed signals from your mic (assuming the mic is connected to those channels). You can set up inputs 1 and 2 as a stereo pair, if using a Telinga Stereo mic.

Before considering the LR mix, I would make sure that all unused channels are disabled - I presume that if you are accidentally applying lots of gain to an empty channel, you could still be adding preamp noise to the mix.

Hello, Jon!

First of all, let me thank you so much for your time, pacience and sharing!

I tryed almost every combinations of gain and levels meters (external knob). Now i use the 32 float, because it is easier... The LR mix is not function in 32 fl. I work only with 1-2 channels. The others are disable and the rec level it´s on too.
My confusion is the fact that for see in the screen the behaviour of sound i have to raise the levels on the external knob at +20/+30db!!! And the result when i normalise the audio is quite bad, almost every time (noises far away appears like cars, motos, plains... And the wind effects are stronger too). So, i give up of see something in the screen and tryed low levels on the external knob. I set them between -17/-22db, depending of using mic ou parabolic. Less that this and the results are not clear enough. So it´s very possible that i miss something important when i set the things to recording. If you be a litlle more pacience, please, listen the 30 sec record i attach, made at 32fl, -20db with the mic. I only normalise it (no filters or noise reduction)... (the wind us about 6/7 km/h and the Oriolus oriolus about 10 metters, more a less). Acceptable? What would you change?

Just as you advise i will convert the 32 float at 24 bit from today!

Thank you, Jon!

 

[Telúrico]

Reading this again I am a bit confused. -3dBfs would mean that the signal should peak on the F6 ballistic meter just below the 0 mark (well half way between the 0 and 6 mark below). I am surprised if you could get near to this with a very low gain level, after all -3dBfs is pretty loud. I would have thought that you would need to have a high recording level (which may create audible noise) to get the meter to peak near -3dBfs. In my thoughts, chasing -3dBfs is more likely to mean you are aware of noise during monitoring, rather than vice versa.

As said in my last post, normalisation is done in post. If your recording is quieter or louder than -3dBfs (and has a good signal to noise ratio) don’t worry, just normalise if afterwards. The EBird requirement should not force your selection of recording level - particularly when you are using 32bit float. The beauty of 32bit float is that it is very forgiving and you can fix things later.

I´m sorry, Jon... The sound looks very low before normalised at -3db. I did not explain myself weell. Sorry! Thank very much!

 

[Jon.Bryant]

My confusion is the fact that for see in the screen the behaviour of sound i have to raise the levels on the external knob at +20/+30db!!! And the result when i normalise the audio is quite bad, almost every time (noises far away appears like cars, motos, plains... And the wind effects are stronger too). So, i give up of see something in the screen and tryed low levels on the external knob. I set them between -17/-22db, depending of using mic ou parabolic.

When you say to 'see in the screen the behaviour or the sound', do you mean to get the balistic meters (the green bars for each channel) to peak at near to -3dB?

Unfortunately I do not know the F6 (and the manual doesn't help much). Assuming the knobs are working as gain controls and not faders, I would have expected the knobs to go from 0dB to something like 70dB. Perhaps someone with knowledge of the F6 can explain. Nevertheless, it seems quite a big jump in your settings to go from plus +20/30dB to -17/-22dB with this control (assuming everything else is set the same). Each 3dB is basically a doubling of the sound level, so you are making a very big change to the recording levels.

You say that 'noises far away appear like cars, motors and planes'. Unfortunately this is a big problem with field recording in general - you need a quiet environment. Shotgun mics like the ME66 are very directional for high frequency noise (such as birds song), but not very good at low frequency noise (including aircraft noise). I think this is to do with the fact that low frequency noise more easily diffracts (bends) around obstacles - a shotgun mic is basically a mic capsule in a diffraction tube, that works by cancelling out sound from the side. The parabolic mic should be more directional and pick up less off axis noise, but low frequency noise can often be diffracted and reflected before it gets to you, so not all sound may be off-axis. If you are using the Twin Science mic with the Telinga, it may be worth positioning the capsule to use the inward facing cardiodal mic, as I think this should pick up less (unfocused) ambient noise.

The problem is that if these sounds are part of the landscape (and captured when you set the recording levels at 20dB), you cannot make them disappear by selecting a different recording level - they are part of the signal to (ambient) noise ratio. One solution is to try to record in the very early morning, when unwanted sound may be lower or absent. I think the only other real solution in a noisy environment is to get as close as possible to the bird. I think that if you halve a distance you increase the sound intensity 4x. This means that if you were 5m from the Golden Oriole (rather than 10), the bird would be four times louder (If you were 2.5m from the oriole it would be 16x louder!). When you normalize the sound level of the Golden Oriole would be the same, but now the unwanted background noise would be 4 x quieter (assuming you have not also got closer to the unwanted noise).

I am wondering how you have managed to get rid of unwanted noise by reducing the gain levels, as this really shouldn't be possible - if you can hear the unwanted noise with some gain settings, it is obviously part of the signal being capture by the mic, so it is magic if you can make it disappear from the recording. Perhaps you are pushing down quite quiet noises below the noise floor for the recording, which would mean that the signal for the unwanted noise would then be lost. However if this is the case you are probably having to amplify the recording a lot in normalisation, including amplification of the noise floor. When I looked at the sonogram for your recording I was a bit surprised that there appeared to be a reasonable amount of general noise - Golden Orioles are quite loud and 10m is not that far away, so I would have perhaps expected a 'whiter background' in the sonogram - perhaps the grey background is the noise floor showing after normalisation?

I wouldn't recommend trying to suppress environmental noises by setting the gain to very low, as this really shouldn't work. I think you need to set the levels to show in the ballistic meters - looking at these (or monitoring the recording) may well help you aim the mic and can be useful. I would go out as early as possible to avoid unwanted ambient noise and also try to get as close as possible to the subject. I would then normalise to a level where you can tolerate any unwanted ambient noise (and forget the -3dB requirement), if it is -6dB or something else so be it. Finally, I would consider post processing - Steinberg Spectral Layers is my recommendation, but it is not that cheap. Using this software, I have found it quite easy to get rid of various unwanted sounds - planes and cars are a still a bit tricky as they generate wide spectrum noise of variable intensity, so not easily got rid of with noise reduction alone.

 

[Telúrico]

When you say to 'see in the screen the behaviour or the sound', do you mean to get the balistic meters (the green bars for each channel) to peak at near to -3dB?

Unfortunately I do not know the F6 (and the manual doesn't help much). Assuming the knobs are working as gain controls and not faders, I would have expected the knobs to go from 0dB to something like 70dB. Perhaps someone with knowledge of the F6 can explain. Nevertheless, it seems quite a big jump in your settings to go from plus +20/30dB to -17/-22dB with this control (assuming everything else is set the same). Each 3dB is basically a doubling of the sound level, so you are making a very big change to the recording levels.

You say that 'noises far away appear like cars, motors and planes'. Unfortunately this is a big problem with field recording in general - you need a quiet environment. Shotgun mics like the ME66 are very directional for high frequency noise (such as birds song), but not very good at low frequency noise (including aircraft noise). I think this is to do with the fact that low frequency noise more easily diffracts (bends) around obstacles - a shotgun mic is basically a mic capsule in a diffraction tube, that works by cancelling out sound from the side. The parabolic mic should be more directional and pick up less off axis noise, but low frequency noise can often be diffracted and reflected before it gets to you, so not all sound may be off-axis. If you are using the Twin Science mic with the Telinga, it may be worth positioning the capsule to use the inward facing cardiodal mic, as I think this should pick up less (unfocused) ambient noise.

The problem is that if these sounds are part of the landscape (and captured when you set the recording levels at 20dB), you cannot make them disappear by selecting a different recording level - they are part of the signal to (ambient) noise ratio. One solution is to try to record in the very early morning, when unwanted sound may be lower or absent. I think the only other real solution in a noisy environment is to get as close as possible to the bird. I think that if you halve a distance you increase the sound intensity 4x. This means that if you were 5m from the Golden Oriole (rather than 10), the bird would be four times louder (If you were 2.5m from the oriole it would be 16x louder!). When you normalize the sound level of the Golden Oriole would be the same, but now the unwanted background noise would be 4 x quieter (assuming you have not also got closer to the unwanted noise).

I am wondering how you have managed to get rid of unwanted noise by reducing the gain levels, as this really shouldn't be possible - if you can hear the unwanted noise with some gain settings, it is obviously part of the signal being capture by the mic, so it is magic if you can make it disappear from the recording. Perhaps you are pushing down quite quiet noises below the noise floor for the recording, which would mean that the signal for the unwanted noise would then be lost. However if this is the case you are probably having to amplify the recording a lot in normalisation, including amplification of the noise floor. When I looked at the sonogram for your recording I was a bit surprised that there appeared to be a reasonable amount of general noise - Golden Orioles are quite loud and 10m is not that far away, so I would have perhaps expected a 'whiter background' in the sonogram - perhaps the grey background is the noise floor showing after normalisation?

I wouldn't recommend trying to suppress environmental noises by setting the gain to very low, as this really shouldn't work. I think you need to set the levels to show in the ballistic meters - looking at these (or monitoring the recording) may well help you aim the mic and can be useful. I would go out as early as possible to avoid unwanted ambient noise and also try to get as close as possible to the subject. I would then normalise to a level where you can tolerate any unwanted ambient noise (and forget the -3dB requirement), if it is -6dB or something else so be it. Finally, I would consider post processing - Steinberg Spectral Layers is my recommendation, but it is not that cheap. Using this software, I have found it quite easy to get rid of various unwanted sounds - planes and cars are a still a bit tricky as they generate wide spectrum noise of variable intensity, so not easily got rid of with noise reduction alone.

Hello, Jon!

Portugal have the bigger relation between kms of roads and people in Europe! And here, in Algarve, the plains are a big problem too! eheh! It´s dificult to record in a quiet places, even on the montains!
I trye to go early in the morning to record, but many times i get frustrated with the noises arround (that Oriolus record was done at 5.30 am). I love getting close to the birds! But species like Oriolus are not a easy task! eheh

I will set the gain like you said (+20/+30bb) to see the green bars on the balistc meters and use a tripod to avoid noise moviments.

Setting the 24 bit, the gain rang goes from +12db to +75db.
32 float, the external knob goes -60 db to +60db.

The noise floor in recording appears, yes, after normalisation. In this case, i don´t know how much is from the wind and how much is from the hiss by recording in a very low gain... But, like you said, it´s a mistake recordin at so low levels and i thank you very much for that! And everything else!
I will see the Steinberg Spectral Layers.

Thank you so very much! I´m learning a lot with you, Jon. If i can do something for you, will be a pleasure!

 

[Telúrico]

When you say to 'see in the screen the behaviour or the sound', do you mean to get the balistic meters (the green bars for each channel) to peak at near to -3dB?

Unfortunately I do not know the F6 (and the manual doesn't help much). Assuming the knobs are working as gain controls and not faders, I would have expected the knobs to go from 0dB to something like 70dB. Perhaps someone with knowledge of the F6 can explain. Nevertheless, it seems quite a big jump in your settings to go from plus +20/30dB to -17/-22dB with this control (assuming everything else is set the same). Each 3dB is basically a doubling of the sound level, so you are making a very big change to the recording levels.

You say that 'noises far away appear like cars, motors and planes'. Unfortunately this is a big problem with field recording in general - you need a quiet environment. Shotgun mics like the ME66 are very directional for high frequency noise (such as birds song), but not very good at low frequency noise (including aircraft noise). I think this is to do with the fact that low frequency noise more easily diffracts (bends) around obstacles - a shotgun mic is basically a mic capsule in a diffraction tube, that works by cancelling out sound from the side. The parabolic mic should be more directional and pick up less off axis noise, but low frequency noise can often be diffracted and reflected before it gets to you, so not all sound may be off-axis. If you are using the Twin Science mic with the Telinga, it may be worth positioning the capsule to use the inward facing cardiodal mic, as I think this should pick up less (unfocused) ambient noise.

The problem is that if these sounds are part of the landscape (and captured when you set the recording levels at 20dB), you cannot make them disappear by selecting a different recording level - they are part of the signal to (ambient) noise ratio. One solution is to try to record in the very early morning, when unwanted sound may be lower or absent. I think the only other real solution in a noisy environment is to get as close as possible to the bird. I think that if you halve a distance you increase the sound intensity 4x. This means that if you were 5m from the Golden Oriole (rather than 10), the bird would be four times louder (If you were 2.5m from the oriole it would be 16x louder!). When you normalize the sound level of the Golden Oriole would be the same, but now the unwanted background noise would be 4 x quieter (assuming you have not also got closer to the unwanted noise).

I am wondering how you have managed to get rid of unwanted noise by reducing the gain levels, as this really shouldn't be possible - if you can hear the unwanted noise with some gain settings, it is obviously part of the signal being capture by the mic, so it is magic if you can make it disappear from the recording. Perhaps you are pushing down quite quiet noises below the noise floor for the recording, which would mean that the signal for the unwanted noise would then be lost. However if this is the case you are probably having to amplify the recording a lot in normalisation, including amplification of the noise floor. When I looked at the sonogram for your recording I was a bit surprised that there appeared to be a reasonable amount of general noise - Golden Orioles are quite loud and 10m is not that far away, so I would have perhaps expected a 'whiter background' in the sonogram - perhaps the grey background is the noise floor showing after normalisation?

I wouldn't recommend trying to suppress environmental noises by setting the gain to very low, as this really shouldn't work. I think you need to set the levels to show in the ballistic meters - looking at these (or monitoring the recording) may well help you aim the mic and can be useful. I would go out as early as possible to avoid unwanted ambient noise and also try to get as close as possible to the subject. I would then normalise to a level where you can tolerate any unwanted ambient noise (and forget the -3dB requirement), if it is -6dB or something else so be it. Finally, I would consider post processing - Steinberg Spectral Layers is my recommendation, but it is not that cheap. Using this software, I have found it quite easy to get rid of various unwanted sounds - planes and cars are a still a bit tricky as they generate wide spectrum noise of variable intensity, so not easily got rid of with noise reduction alone.

Just one more thng, Jon...
One day i was with a sound technician (for a short time), and i said to me that was strange the green bars responding at so high levels (+20/30db), but i didn´t find why (never worked in the zoom f6). And tell me something else: to work with 24 bit (not withe the 32 float) and to recording close birds, set the trim above +20db and bellow 0 db on the external knob... i tried that... no green bars on the screen to see )and the sounds were not good enought)! eheh! Honestly, before talking to you i was pretty much lost!

 

[Jon.Bryant]

I am not sure I have really solved your problems/questions. What you really need is someone who uses the Zoom F6 to explain exactly how the recorder works.

When you say the 'green bars responding' what do you mean? Do you mean you don't see anything, or just that the bars are quite small? It would be quite strange if you don't see anything at all.

It also seems a bit on/off to either select -17/-22 to +20dB. Can you confirm if the meters respond to you slowly turning the track knob? What do you see at +10 or +15dB? I would hope that the bars slowly increase in size as you turn the knob. If the bars don't increase in size, then this would suggest the level is being monitored before the gain/fader is applied, but then this would not explain why you can then see the levels if you turn up the gain/fader to +20dB.

I understand that the track trim you mention is a defined gain level that is applied to 16 and 24 bit recordings. Something I read suggests that the recorder works in 32bit float (with set gain levels) and if required converts from this recording to a 16 or 24bit recording with the defined trim applied. This same article suggested that there is no control over the gain levels for the 32bit track which is set at a default level optimised for the AD converters - this appears somewhat different to what the manual suggests.

When you say that you tried 24bit with trim at +20dB and gain below zero, but there were no green bars on the screen, agian what do you mean? Do you mean no bars of just short ones? Did you try turning the track knob to above 0dB or changing the trim to a higher level, and if so did you see anything?

If in you can play with trim and gain levels to get an acceptable 24bit recording, then it may be worth having a play with dual mode (recording a 32bit and 24 bit version of the same track simultaneously). The article I read, suggests that in this mode you can control the gain of the 24bit track (so you can learn the craft of setting gain levels), but that you have a 'rescue' 32bit track at a default gain level, in case you get the 24 bit recording levels wrong.

 

[Telúrico]

Hello, Jon!

For all i can see, with 32 float, you van not control the gain, only the levels on the knob. The trim don´t even appears available.

When i was recording the Oriolus the green bars appears at +10, but very very small. Bellow that no sign on the green bars. So, yes if i increase the levels in the knob slowly i see the bars increase too. To put the green bars reaching -12db, like eBird and other tutorials say is the right thing to do, with 32 float that means more a less, depends on the bird distance, +20 to +30bd or more (depending of how strong sings the bird).
The same with 24 bit, trim +20 db and -1db in the knob. Nothing of green bars and i have to push the levels on the knob to +20/+30db or more!

Before reading the tutorials i was recording with the trim at +37/+45 db, sometimes +50 db (too high) and the levels on the knob between -7 and -1db. And work only with -16db normalisation, the bass, trebel and volume, but in some recordings i used only with the bass, trebel and volume. I got more white sonograms then! But even on +45/+50 db of trim the greens bars appears reaching -12db only at, minimum, +10 db on the knob levels.

I will try that dual (24/32). One day i will find the right balance! eheh! With your hints, i will try some combinations. However, and that´s why i thing something is wrong wiht the records setting, i already experiment at home, for example; 32 fl and +20db; 32fl and 0db; 32 fl and -12db, I didn´t notice no difference considerable, for not to say any, after normalisation! Quite bewildering!

You said that the normalisation at -3db is not an obrigation, but is the correct thing to do when you get a good recording? You use it? To how many db´s? Do you use bass, trevel and volume, for example?


Thank you!

 

[Jon.Bryant]

Hmm. I felt like I was having Deja Vu - well here is the previous thread about the F6 recording levels Trim and Gain settings.

Hopefully this may help a bit, but also I note that some participants in the thread have F6 records, so probably worth opening up a conversation direct if you are struggling.

One thing that I think I got wrong(ish) in the previous thread is that I think that the recorder works all the time in 32bit float, so if you want a 24 or 16bit recording it converts from 32bit float in real time. I understand that the gain is optimized for 32bit float, but this may be too low for 24 or 16 bit, hence you can apply trim to correct this - but only in 16 or 24 bit recordings (trim is not applicable and will not work in 32bit float).

You say

you can not control the gain, only the levels on the knob

My understanding is that these knobs can be set to work in two ways - the default is gain control prior to a recording, and faders during a recording (with the latter only changing the LR mix). The block diagram for float mode in the referenced thread, states that you can change this default to gain control prior to and during a recording. So either way I think you do have gain control! Be careful though, as I think you can monitor pre and post faders (I am not sure if this also applies to the meters). If the meters can be set to pre-faders and you have the knobs working as faders, then twiddling them during a recording will not make any difference! You would have to set monitoring to post faders. I am not sure if monitoring in float treats the dual function of the knobs (gain and faders) the same. I was thinking that pre-gain/fader monitoring may be your problem, but in this case, I wouldn't expect to see any change in levels no matter how far you turned the knob.

Do you use bass, trebel and volume, for example?

I am not sure what you mean here. Basically normalisation should be a process to detect the loudest element of the recording and then to apply gain to the whole recording, so that the loudest element reaches the specified value. Normalisation is not normally selective on frequency (i.e. bass, treble etc.). What software are you using?

With the free software Audactity, the menu option is Effect>Volume and Compression>Normalize. You then get a dialogue with some check boxes - The 'Remove DC offset' should be checked, the 'normalize peak amplitude to' should be checked and you should set the value to whatever you want (say -3dB), and the 'Normalise stereo channels independently' should be unchecked (but not relevant to mono in any case).

You said that the normalisation at -3db is not an obligation, but is the correct thing to do when you get a good recording?

It should be OK, but try and then listen to the normalised recording. If the normalisation process has lifted the background too much, then try again with a lower value (or if you have editing software, see if you can then reduce the background noise in post - but this is not easy and you can end up with an unnatural sound if you over edit).

 

[Telúrico]

Hmm. I felt like I was having Deja Vu - well here is the previous thread about the F6 recording levels Trim and Gain settings.

Hopefully this may help a bit, but also I note that some participants in the thread have F6 records, so probably worth opening up a conversation direct if you are struggling.

One thing that I think I got wrong(ish) in the previous thread is that I think that the recorder works all the time in 32bit float, so if you want a 24 or 16bit recording it converts from 32bit float in real time. I understand that the gain is optimized for 32bit float, but this may be too low for 24 or 16 bit, hence you can apply trim to correct this - but only in 16 or 24 bit recordings (trim is not applicable and will not work in 32bit float).

You say

My understanding is that these knobs can be set to work in two ways - the default is gain control prior to a recording, and faders during a recording (with the latter only changing the LR mix). The block diagram for float mode in the referenced thread, states that you can change this default to gain control prior to and during a recording. So either way I think you do have gain control! Be careful though, as I think you can monitor pre and post faders (I am not sure if this also applies to the meters). If the meters can be set to pre-faders and you have the knobs working as faders, then twiddling them during a recording will not make any difference! You would have to set monitoring to post faders. I am not sure if monitoring in float treats the dual function of the knobs (gain and faders) the same. I was thinking that pre-gain/fader monitoring may be your problem, but in this case, I wouldn't expect to see any change in levels no matter how far you turned the knob.


I am not sure what you mean here. Basically normalisation should be a process to detect the loudest element of the recording and then to apply gain to the whole recording, so that the loudest element reaches the specified value. Normalisation is not normally selective on frequency (i.e. bass, treble etc.). What software are you using?

With the free software Audactity, the menu option is Effect>Volume and Compression>Normalize. You then get a dialogue with some check boxes - The 'Remove DC offset' should be checked, the 'normalize peak amplitude to' should be checked and you should set the value to whatever you want (say -3dB), and the 'Normalise stereo channels independently' should be unchecked (but not relevant to mono in any case).



It should be OK, but try and then listen to the normalised recording. If the normalisation process has lifted the background too much, then try again with a lower value (or if you have editing software, see if you can then reduce the background noise in post - but this is not easy and you can end up with an unnatural sound if you over edit).

Hello, Jon!

I ask for support to zoom f6. I am waiting for a response...

I try to not over edit. Use the normalisation and some times de "bass and trebel", from effects audacity menu.

I was testing... At 32 float i can see difference in the 1 and 2 channel bars when i work with the knob and also at the LR green bars. When i set dual (32 float and 24) i can´t see no difference at the 1 and 2 channels green bars when i use the knob. Only in LR green bars and quite a difference. In both cases the LR/fader was off. I the dual mode, the only way to control the green bars is with the trim (higher trim, bigger bars and the opposite).

I only see pre and post fader controls in the "output" menu - Line out - routing, here i can set the 1 and 2 channels to pre or pos fader. Can set the level too in output menu - line out - level - LR/Fader (is at -3db) and Line out (also at -3db). In rec menu - LR track i can set LR Fader and Line out (both -3db), and i can too set on or off the LR. I don´t know if this make any difference in the recording. Frankly i don´t know what fader is exactly. I need to study some sound notion deeper.

Thank for your time and care, Jon! I will keep trying! And i will try to talk with someone how take part of the "gain and trim settings" coversation. Thank you!

 

[Telúrico]

Hello, Jon!

I ask for support to zoom f6. I am waiting for a response...

I try to not over edit. Use the normalisation and some times de "bass and trebel", from effects audacity menu.

I was testing... At 32 float i can see difference in the 1 and 2 channel bars when i work with the knob and also at the LR green bars. When i set dual (32 float and 24) i can´t see no difference at the 1 and 2 channels green bars when i use the knob. Only in LR green bars and quite a difference. In both cases the LR/fader was off. I the dual mode, the only way to control the green bars is with the trim (higher trim, bigger bars and the opposite).

I only see pre and post fader controls in the "output" menu - Line out - routing, here i can set the 1 and 2 channels to pre or pos fader. Can set the level too in output menu - line out - level - LR/Fader (is at -3db) and Line out (also at -3db). In rec menu - LR track i can set LR Fader and Line out (both -3db), and i can too set on or off the LR. I don´t know if this make any difference in the recording. Frankly i don´t know what fader is exactly. I need to study some sound notion deeper.

Thank for your time and care, Jon! I will keep trying! And i will try to talk with someone how take part of the "gain and trim settings" coversation. Thank you!

well, i read the "gain and trim settings"... and understand that the 1-2 channels don´t have fader control, as you said there. Very useful, thank you!