1.5x oversampling

[Nick]

I was wondering if it is feasible to have oversampling at 1.5. my dac won't do 192. So my 96 files remain 96 whereas my 88 files happily play at 176. Likewise when i attempt 196 the player brilliantly outputs at 144 rather than 96 like other players. This got me curious as to whether my 96khz files can be oversampled to 144khz(1.5x)? By default I have oversampling set to 4x and congratulate the developers on how well it deals with the occasions 4x is not possible due to endpoint limitations.

[blaubär]

As to oversampling the developer wrote :

Well, setting high sample rate is also oversampling but Oversample option will work a bit differently, it will sent higher frequency which is multiple to the assigned ratio 2x, 4x, ... e.g. 44100 file with 2x will be played with 88200 Hz frequency whereas when user set some fixed frequency, for example 96000 then 44100 will have to be resampled with 2.17687 rate which is not even and degrades quality. So Oversample option automates it and makes sure that resampling frequency is even to the frequency of the file.

Also see :

Also, being a complete Noob to HiRes, just playing around with all the settings and frequencies...

You can make an argument for oversampling non-high-res-audio, that will increase the sampling rate before the DSP ( digital signal processing = equalizer crossfeed compressor etc ), that might optimize the calculations there. Again I don't claim to hear any difference.

And you can make an argument for upsampling not oversampled non-high-res-audio before the DAC ( digital analog converter, the thing where neutron delivers its output to ), because the DAC would upsample anyway and neutron's upsampling might be better than that of the DAC ... I find this argument a bit dubious.

Some things about the sampling rate that I found out in the meantime :

1. If Neutron resamples, it resamples the output of the decoder and feeds the DSP with the new sampling rate, so the DSP will work with the new sampling rate, and deliver data with this rate to the output.

2. If the firmware resamples, it will do so only after the DSP and only if it doesn't support the sampling rate delivered by the DSP. If the DSP delivers a supported sampling rate the firmware won't resample.

3. If you activate Audio Hardware > Oversampling then Audio Hardware > Frequency will be ignored.

4. Audio Hardware > Oversampling sets the sampling rate to an even multiple of the input. This is a good thing on its own, better than a non-even multiple. But if the resulting sampling rate isn't supported by the device the firmware will resample again after the DSP to the best suitable rate.

5. Audio Hardware > Frequency will set the sampling rate to a value supported by the device. This is a good thing on its own. But it might force Neutron to resample to an non-even multiple, which isn't as perfect as resampling to an even multiple.

So Oversampling won't resample with a factor of 1.5 .

[Nick]

Ok. I thought as it can downsample at 0.75 it could oversample using the same fractal method.

[blaubär]

Ok. I thought as it can downsample at 0.75 it could oversample using the same fractal method.

It can do that. That's then oversampling in the general sense of upsampling. Neutron uses the term "oversampling" in a narrower sense, as described above. If your DAC supports a sampling rate you can choose it. That means if you play a song with a sampling rate of 44.1 kHz and you want to render and play it with 88.2 kHz and your DAC supports this then you can either explicitly choose the 88.2 kHz in Audio Hardware > Frequency or you can activate Audio Hardware > Oversampling > Ratio: 2x, this both leads to the same : before the DSP the sampling rate will be doubled and the DAC will get that new sampling rate. If you want to use 96 kHz and your DAC supports this then you can set that frequency directly, but not with oversampling. Unless your song has a sampling rate of 48 kHz, then it's the other way round. And so forth ...