Barbarella
New Member
Hi there! I need to know if OBS uses 16 or 24 Bit for audio?? I want to match all of my other devices. Thanks!
Question / Help help: 16 or 24 Bit??
- Thread starter Barbarella
- Start date
koala
Active Member
That's a bit difficult to say. It's not documented. Browsing in the source seems to reveal that OBS will autodetect and use whatever format is used by the corresponding device(s).
If the devices you use with OBS all have the same audio format, OBS will use the same and never resample.
If it's necessary to convert (resample) because a source device uses a different format than a destination device, it tells the resampler function the source format and the destination format and let the resampler do what's necessary to convert.
So the question which audio format OBS will use isn't relevant really. To avoid resampling, use the same format everywhere, and OBS will respect this automatically.
The only thing you need to set in OBS is the sampling rate - this should indeed match the sampling rate of all your devices.
If the devices you use with OBS all have the same audio format, OBS will use the same and never resample.
If it's necessary to convert (resample) because a source device uses a different format than a destination device, it tells the resampler function the source format and the destination format and let the resampler do what's necessary to convert.
So the question which audio format OBS will use isn't relevant really. To avoid resampling, use the same format everywhere, and OBS will respect this automatically.
The only thing you need to set in OBS is the sampling rate - this should indeed match the sampling rate of all your devices.
Barbarella
New Member
ChorizoNinja
New Member
@koala @Barbella I know this is an old post, but I would like to share some things.
OBS captures audio at 32 bit. I know this for a fact as the api will spit audio data in floating point (32bit). This is done because your output audio of your DAC can be configured to something "stupid" like 192Khz/32bits PCM. If you play some file in that format with that sample rate/ bit rate, obs will resample it to 44.1/48khz and 16bit. The reason for this is simple, you don't need more range for broadcasting/listening.
44.1Khz and 48Khz offers enough granularity of frequencies, being 22050Hz and 24000Hz, well beyond human hearing. Going beyond that would only improve ultrasonic harmonics that "might" affect audible frequencies if you have golden trained ears but, for the average joe, including myself, I can't hear a difference, As well, for those harmonics to be even present, you will need a really good pair of headphones that display ultrasounds... so there is that.
Now talking about bit depth. Obs will output 16bit, and to be honest, 16bit is more than enough, and here is why.
For any bit depth in a fixed point scale (PCM), you can apply the following formula:
20*log10(1/MAX_BIT_VALUE)
16bit offers you 96db of dynamic range. DBFS scale would go from -96db to 0db. The reason for this is really simple, 16bit can represent 65,536 different values. If you want to hear the quietest sound in a room with 25 db (DB SPL) of noise floor, the loudest sound will blast your ears at 120db (DB SPL), being really painful. So lets say, this is plenty of dynamic range. So trust me when I say, -96db is as dark of a noise floor at any humanly tolerated listening levels.
If you apply the same for 24bit (16,777,216) and 32bit(16,777,216), you will end up with 144db of dynamic range (-144 to 0db DBFS) and 192db (-192 to 0db DBFS) respectively. If the previous one is painful, this will knock you out, your ears will be bleeding and most likely you will be deaf. Those are stun grenade levels of sound pressure.
24bit and 32bit PCM and 32bit float point audio (this one has a stupid dynamic range of 1528db, from -758db to 770db in the DBFS scale, and yes, it goes below 0), is used mainly for mastering audio, so you can play with gain and different effects without audio clipping. I hope this makes sense.
OBS captures audio at 32 bit. I know this for a fact as the api will spit audio data in floating point (32bit). This is done because your output audio of your DAC can be configured to something "stupid" like 192Khz/32bits PCM. If you play some file in that format with that sample rate/ bit rate, obs will resample it to 44.1/48khz and 16bit. The reason for this is simple, you don't need more range for broadcasting/listening.
44.1Khz and 48Khz offers enough granularity of frequencies, being 22050Hz and 24000Hz, well beyond human hearing. Going beyond that would only improve ultrasonic harmonics that "might" affect audible frequencies if you have golden trained ears but, for the average joe, including myself, I can't hear a difference, As well, for those harmonics to be even present, you will need a really good pair of headphones that display ultrasounds... so there is that.
Now talking about bit depth. Obs will output 16bit, and to be honest, 16bit is more than enough, and here is why.
For any bit depth in a fixed point scale (PCM), you can apply the following formula:
20*log10(1/MAX_BIT_VALUE)
16bit offers you 96db of dynamic range. DBFS scale would go from -96db to 0db. The reason for this is really simple, 16bit can represent 65,536 different values. If you want to hear the quietest sound in a room with 25 db (DB SPL) of noise floor, the loudest sound will blast your ears at 120db (DB SPL), being really painful. So lets say, this is plenty of dynamic range. So trust me when I say, -96db is as dark of a noise floor at any humanly tolerated listening levels.
If you apply the same for 24bit (16,777,216) and 32bit(16,777,216), you will end up with 144db of dynamic range (-144 to 0db DBFS) and 192db (-192 to 0db DBFS) respectively. If the previous one is painful, this will knock you out, your ears will be bleeding and most likely you will be deaf. Those are stun grenade levels of sound pressure.
24bit and 32bit PCM and 32bit float point audio (this one has a stupid dynamic range of 1528db, from -758db to 770db in the DBFS scale, and yes, it goes below 0), is used mainly for mastering audio, so you can play with gain and different effects without audio clipping. I hope this makes sense.
mfhansen1982
New Member
I think you meant 2^32 = 4,294,967,296 (32 bit)
Nevertheless, here is a question I am wondering about. I am using a custom ffmpeg setup where I am outputting PCM audio into a mkv file (which I then remux to mov) at 24 bit depth. My Scarlett 2i2 is able to output 48 kHz @ 24 bit. Will OBS then resample to 48 kHz @ 16 bit and then put that into a 24 bit PCM track or do I actually get the full 24 bit audio.
I am doing a lot of audio editing with my tracks, so I would like to retain the highest possible bit depth that my device can offer (which is 24 bit).
Cheers,
Michael
DayGeckoArt
Member
There's another reason to 96khz or 192khz besides golden ears-- Most content is 44khz now but much is still 48khz. The video standard is 48khz so really we should be recording and streaming in 48. Either way, if your soundcard doesn't match, sound is constantly resampled so you get a reduction in quality. Just having everything resample up to a much higher sample rate, 96khz or 192khz, is better IMO than listening to 44khz content at 48khz
richardpohl
New Member
I would like to also know, if just record using these options with my Audient id44, is everything getting resampled by OBS? It is not a true 24bit recording then?