Mark has been helpful with PM. I now have several paths to investigate. I will post those results when I have them.
Next steps for me.
1. Set up a blind listening test protocol
2. Compare Minisharc and bypass with just one channel playing at a time.
3. Jitter measurement.
It's quite easy to set up a blind test. This file might helpIn addition to what Sy has to say, I have found that it may be necessary to consciously identify and memorize the sound of small differences when subject to blind testing. Parts of the brain that recognize words in a language already mastered aren't so good an picking out small differences in regional accents in a language that is not known by the listener. The differences are still there nonetheless, perhaps easy to hear when relaxed and the brain is processing sound without distraction. Test taking may not seem like it should be much of a distraction, just pushing a button or something, but it can be distracting in reality. Holding peculiar sounding small differences in memory may be like memorizing several digits, say, 6-10 digits while doing something else, say, thinking about whether x is b and y is a or x is a and y is b. Try it sometime. It can take some practice to get used to.
IMHO, some people have been fooled into thinking the can't hear small differences as a result of participation in inappropriately conducted blind listening tests.
Also, what is audible depends on the whole reproduction system. Masking can occur due to distortion and noise in the system, which can happen even if it measures well due to the presence of nonstationary or quasi-nonstationary distortion and noise. My advice: use good reproduction equipment if possible.
No surprise if others disagree with my comments above. Just my two cents.
IMHO, some people have been fooled into thinking the can't hear small differences as a result of participation in inappropriately conducted blind listening tests.
Also, what is audible depends on the whole reproduction system. Masking can occur due to distortion and noise in the system, which can happen even if it measures well due to the presence of nonstationary or quasi-nonstationary distortion and noise. My advice: use good reproduction equipment if possible.
No surprise if others disagree with my comments above. Just my two cents.
Last edited:
Are you implying that subjects(including this case) outside peer-reviewing process do not exist?
JD,
Old thread, but we experienced the same, and after investigations we concluded its the ASRC... It sounds really poor and one reason why I avoid DSP... if you cannot get ASRC to sound transparent, what chances is there for more complex DSP functions?
Eh? Thread is less than a week old.
Now 'It sounds really poor' is not particularly helpful, esp as its at best a teeny change. I don't suppose there were any measurements. As for 'what chances is there for more complex DSP functions' that just appears to be you showing a fear of mathematics rather than any reasoned contribution.
Now 'It sounds really poor' is not particularly helpful, esp as its at best a teeny change. I don't suppose there were any measurements. As for 'what chances is there for more complex DSP functions' that just appears to be you showing a fear of mathematics rather than any reasoned contribution.
To be fair to AD, the ASRC is effectively a fixed function whereas the rest of the chip is a blank sheet. It is for you to get it to do whatever you want.
Last edited:
On the miniDSP forum the OP asked the same question and provided one detail that I see left out of this discussion, he is not using a miniSHARC alone he is using a miniSHARC with a DIGI-FP.
In this configuration the ASRC is performed by the SRC4382 based DIGI-FP and NOT the ADSP21369 (although of course the clock is still being provided by the miniSHARC).
Michael
In this configuration the ASRC is performed by the SRC4382 based DIGI-FP and NOT the ADSP21369 (although of course the clock is still being provided by the miniSHARC).
Michael
Thanks for the comment. Do you mind sharing how you isolated the coloration cause to the ASRC?
I assume we only need ASRC if were running the source and DSP with different MCLK's correct?
jderimig,
You might want to take a look at another thread to help with ASRC questions: Asynchronous Sample Rate Conversion
You might want to take a look at another thread to help with ASRC questions: Asynchronous Sample Rate Conversion
Thanks for the thread link, good stuff!
But to confirm my simple question, if my source and DSP use the same MCLK then then I should be able to configure the DSP to be synchronous with Fs and not need the ASRC stage correct?
Example with similar setup to mine:
IanCanada's FiFoPI -> I2S -> DSP, use common oscillator for FiFoPI and DSP
???
A FIFO can be used instead of an ASRC if the nominal input and output sample rates are to be the same. In that case master clocks can be different so long as the FIFO has enough buffer memory to manage the difference between exact clock frequencies. Also, a DSP could be used before or after the FIFO and no ASRC would be needed for jitter attenuation. Probably better to put the DSP before the FIFO in that case.
On the other hand, if the intent is to change sample rates using the DSP, and if the master clock is the same for both sample rate domains, then synchronous sample rate conversion could be used instead of asynchronous. As in the first paragraph above, if there is a FIFO the DSP could come before or after it.
On the other hand, if the intent is to change sample rates using the DSP, and if the master clock is the same for both sample rate domains, then synchronous sample rate conversion could be used instead of asynchronous. As in the first paragraph above, if there is a FIFO the DSP could come before or after it.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.