Topic: reducing hiss on playback using an sbm-1?

[papajohn]

So, thanks to Covid and safer at home protocols,  I've been transferring my old cassette live music collection.  Currently I am using a cassette deck > analog cable > Marantz PMD661 at 24/48 & 24/96.  Although the levels can be adjusted on the Marantz, most of the time the results are still low levels.   Would it be preferable to continue using Sound Forge to increase levels after the transfer and get more bits at a higher rate or would it be preferable to place an SBM-1 before the Marantz but only get 16/48?   So I suppose my question is will the SBM-1 reduce the hiss and noise by allowing a higher volume of music to be passed through to the Marantz?  Your opinions are appreciated.

[goodcooker]

If it's "hissy" adding gain at any stage will amplify the noise along with the program material.

Try using a noise reduction plugin in Soundforge to take as much noise out of the file as you can at a lower volume level then amplify after noise reduction.

[jerryfreak]

So, thanks to Covid and safer at home protocols,  I've been transferring my old cassette live music collection.  Currently I am using a cassette deck > analog cable > Marantz PMD661 at 24/48 & 24/96.  Although the levels can be adjusted on the Marantz, most of the time the results are still low levels.   Would it be preferable to continue using Sound Forge to increase levels after the transfer and get more bits at a higher rate or would it be preferable to place an SBM-1 before the Marantz but only get 16/48?   So I suppose my question is will the SBM-1 reduce the hiss and noise by allowing a higher volume of music to be passed through to the Marantz?  Your opinions are appreciated.

the dynamic range of your marantz will exceeds the dynamic range of your cassette (even at 16/44). as long as the levels are peaking above  -30dB you are capturing all of the cassette material. There is nothing you can really add in the middle of that chain to reduce the tape noise relative to the source material, as goodcooker said, you'd need to remove it digitally in a DAW. And that typically takes a little of the high end with it.

plugins are improving all the time so if youre gonna mess with noise reduction, save the original unmolested transfer so you can give it another go down the road if desired. If you do it enough times you will find that you get better at it, even with the same tools.

[kuba e]

I also think that 16 bits is enough for cassette transfer. If you are using a noise removal plugin, it may be worth considering to do transfer with a higher sample rate e.g. 96 kHz. In a case you are not limited by storage space. Here is a thread where the theory about bit depth and sample rate was explained by our friends (it is long reading):
https://taperssection.com/index.php?topic=195295.0

[papajohn]

thanks for the replies.  your guidance is appreciated.

[Chilly Brioschi]

Hiss is a known quantity, but what types of noise do you want to remove?
I rember cassettes, there are many... wow and flutter of course, but do you mean mic handling noises, audience wooo, clapping ?

[Jammin72]

While it may be an extra step, doing the transfer in 24 bit will give you more headroom if you do want to boost a lower level recording in post.  16 bit IS enough but who wants to run through a whole tape to find the peaks before setting the recording level appropriately?   Everyone is right though hiss is a different issue.

[jerryfreak]

While it may be an extra step, doing the transfer in 24 bit will give you more headroom if you do want to boost a lower level recording in post.

theres no headroom to be had when transferring a recording with 50-60 dB of dynamic range to a 16-bit container with 96 dB of dynamic range. 144dB range of 24 bit just adds 48 more dB of purely noise

any cassette is completely saturated by +10dBU. by setting your levels on your 16-bit AD between -20 and -12dbU where your cassette is at 0dBU, thats about as good as you can do to capture the full dynamic range of the cassette.

[Gutbucket]

any cassette is completely saturated by +10dBU. by setting your levels on your 16-bit AD between -20 and -12dbU where your cassette is at 0dBU, thats about as good as you can do to capture the full dynamic range of the cassette.

This.  16 bits is plenty.  Setting levels set as indicated above is the key. 

..and represents good practice either way, because if you record to 24 bit based on the understanding that doing so will allow you to set recording levels lower, you could conceivably run into the EIN noisefloor of the 24bit recorder or interface's analog front end if levels are low enough.  Recording to 24bit will most likely provide only a somewhat greater overall dynamic range equating to something between 17 to 20 bits total, certainly not the full 24.  If you don't know the EIN of the digital recording device, this is unknown and you should set levels as indicated above regardless.

[morst]

if you ever wanted to use advanced noise removal in the future, having a high resolution version of that noise might be useful

[Jammin72]

Thanks for the clarification on the headroom.

[kuba e]

if you ever wanted to use advanced noise removal in the future, having a high resolution version of that noise might be useful

This should apply to the sampling frequency when we are planning to process the signal digitally. I am not sure if this should apply to the bit rate. Additional bits will be only zeros The lower bits will contain only noise. I'm not sure if we can restore the original signal from the noise. I don't think that's possible.

I hope the picture is not misleading.

[morst]

This should apply to the sampling frequency when we are planning to process the signal digitally. I am not sure if this should apply to the bit rate. Additional bits will be only zeros The lower bits will contain only noise. I'm not sure if we can restore the original signal from the noise. I don't think that's possible.

I have little idea what will be invented, but I know that if the data don't exist, you won't have a chance to decide whether or not to throw it away later!?
I'm most interested in noise within the human hearing spectrum, but it might be the case that high frequency analysis could help remove audio spectrum artifacts.
What is there was some high pitched trigger signal just prior to each instance of an intermittent noise, for example?

[kuba e]

I thought from the opposite point of view. When the noise exceeds and overlays the original signal, there is no more data to be recovered from these low levels. But I'm not a noise specialist. It probably depends on the type of noise. These are interesting things.

I prefer 16bit due to capacity and easier handling of files. Even for 16bit, the quantization noise is several times smaller than the noise of an old cassette. Probably more important is to set the right recording levels as Gutbucket explained. But I agree with you, If I had a very rare recording, then I would transfer it to 24bit.