Tell it to Leo, not me. He was the one who said it in his video.
Sonicake NAM Captures?
- Thread starter pipelineaudio
- Start date
DLC86
Roadie
- Messages
- 803
SRC
Mike Oliphant
Groupie
- Messages
- 69
This particular issues isn't caused by the hardware being 44.1k - it is caused by software taking a NAM model trained on 48k data and interpreting it at 44.1k. This should be solvable by doing resampling during the capture process.
Sascha Franck
Rock Star
- Messages
- 7,474
Sample rate conversion should be very little of an issue these days. I think it's possibly rather *not* converting it but playing it as if it was a 44.1k file when it's actually a 48k file. Possibly sort of comparable (if not even remotely as noticeable) to playing back a 48k file in a DAW set to 44.1k.
Yeah, likely this.
pipelineaudio
Shredder
- Messages
- 2,139
Why did the dimehead NAM player do so bad?
Sascha Franck
Rock Star
- Messages
- 7,474
Whut?
- Messages
- 2,039
97% Doug?
pipelineaudio
Shredder
- Messages
- 2,139
On Leo's video it only got to -40db in the null
DLC86
Roadie
- Messages
- 803
I don't think that's the best way to solve the issue though, cuz you're forced to only use 44.1 kHz models that way. The best would be what the nam plugin does: read the original sample rate of the capture (which is stored in the metadata) and apply SRC pre and post to match the sample rate of the interface (or the pedal in this case)
Afaik you can't convert the sample rate of an AI generated model, what you can do is apply SRC pre and post (what nam and even tonex do).
Mike Oliphant
Groupie
- Messages
- 69
By "resampling during the capture process" I meant the process of the device capturing the NAM model into its own internal format, not the original NAM capture process.
The lower the number, the better result.
-40 is way better then -27.
Sascha Franck
Rock Star
- Messages
- 7,474
Why would that be "only"? -40dB is pretty good already. Lower anything you listen to by 40dB and what's left is the level range of difference, so to say. I know, it doesn't 100% work that way, but still, -40 is heading into the "indiscernable for the human listener" direction already.
Fwiw, as far as Leo Gibson's Kemper comparison goes, I think it's not as valid as the Kemper is using quite a different approach and the outcome can be pretty different depending on the amp you've profiled. As the Kemper is using baseline amps as a foundation (plus some wicked match EQ trickery), when you try to profile an amp not close to any of those baseline amps, the result will be less authentic. Which is also why the Kemper doesn't do well in profiling amps with "complexed" drive pedals in front, whereas NAM has way less issues doing so.
Sascha Franck
Rock Star
- Messages
- 7,474
Ah ok. I actually have no idea how any of the internals work.
pipelineaudio
Shredder
- Messages
- 2,139
-40 is GREAT for most types of audibility tests, but in this case, he had stuff before getting it below audibility in general, like -90 to -120
pipelineaudio
Shredder
- Messages
- 2,139
Ahh ok, that makes sense...I gotta watch that video closer
Well.... it's all me, BUT there is less of me, so...97% Doug?
budgetguitarshow
Newbie
- Messages
- 12
Posting a video from your channel without engaging in the conversation topic is borderline spammer behaviour
2dor
Shredder
- Messages
- 1,115
I think they were backordered for a good while so sales seemed to have been fine.
It is a bit on the pricier side for me. Thar and not being able to double as an Audio interface, the input calibration and file management were what kept me from personally pulling the trigger on it. Most of the other stuff is alright for what the unit wants to be.
Lysander
Shredder
- Messages
- 2,477
I'm really tempted to snatch one of these - and i don't care much about captures to begin with.
Looks like a great piece of practice kit at ~$50; i like how everything can be edited on the device itself.
Similar threads
