Phono Cartridge Response Measurement Script

[Quando]

4. Open the script in Spyder and customize with minor edits
??????

 

[Quando]

 

[Thomas_A]

Unless it is written in a clear, simple and readable way, it is not worth wasting time.

Well, I give you what is written, what you find is worth spendiing time on or not is not my task to judge.

 

[Quando]

Well, I give you what is written, what you find is worth spendiing time on or not is not my task to judge.

I started Spyder and I don't understand what to do next

 

[Thomas_A]

I started Spyder and I don't understand what to do next

Someone else will answer. I use MacOS.

 

[Quando]

Which test signals from the Clerauadio CA-TSR-107 board should I copy to the computer as:
wavefile1.wav
wavefile2.wav

 

[narud]

View attachment 397508

Which test signals from the Clerauadio CA-TSR-107 board should I copy to the computer as:
wavefile1.wav
wavefile2.wav

track 1 (left channel sweep) is file_0. track 2 (right channel sweep) is file_1. on the right channel sweep, reverse the channels so the sweep is in the left channel before running the script.

 

[Quando]

track 1 (left channel sweep) is file_0. track 2 (right channel sweep) is file_1. on the right channel sweep, reverse the channels so the sweep is in the left channel before running the script.

I recorded track 1 and track 2 in one take.
What next?

I understand that I have to cut out the Pilot signal from both tracks
Split the recording into file1 and file2
I don't understand the issue of reversing the channels?

 

[narud]

I recorded track 1 and track 2 in one take.
What next?

I understand that I have to cut out the Pilot signal from both tracks
Split the recording into file1 and file2
I don't understand the issue of reversing the channels?

select each sweep only, with no silence after them, and export them as separate files. both the left and right channel sweeps have to be in the left channel. thats how you get the crosstalk measurement. also, if you record 16bit 44.1, you arent going to get distortion out to 20k.

 

[Quando]

select each sweep only, with no silence after them, and export them as separate files. both the left and right channel sweeps have to be in the left channel. thats how you get the crosstalk measurement. also, if you record 16bit 44.1, you arent going to get distortion out to 20k.

Looking at my chart it looks terrible

 

[Thomas_A]

Looks a bit hot in the treble. Best to do it in 96 kHz as mentioned; distortion plots will be better and not cut off.

 

[Quando]

Looks a bit hot in the treble.

This is the sound of Eskimos

 

[Watire]

track 1 (left channel sweep) is file_0. track 2 (right channel sweep) is file_1. on the right channel sweep, reverse the channels so the sweep is in the left channel before running the script.

This is a piece of information that totally eluded me. Now I understand why the script never seemed to work !
We do need a step by step tutorial in one post. Obviously also on the github page.
Thanks a lot, now I will be able to contribute, @Quando and @narud you have made my day.

 

[Quando]

Great script!

However, I suggest adding the 0-20 Hz range to the measurement as well
There are so many different, generally stupid, guides on the Internet on how to match the cartridge compliance to the effective mass that people do it wrong, or do not do it at all.

It would be nice to see what the amplitude level of the unwanted resonance is, for example +1 dB or maybe +12 dB, (relative to the level of 1 kHz, on the graph)

 

[USER]

However, I suggest adding the 0-20 Hz range to the measurement as well

The problem is that you can't make up something that is not there. The test record sweeps are of a specific range of frequencies. The issue goes both ways, with the test records going beyond 20kHz (such as the JVC-TRS-1005) having their own set of problems, not to mention being nearly impossible to find these days. Simply, we are limited to what the test records afford us. If you were following 5 years ago you would have seen that the script initially had a range from 1-20kHz (for very good reasons), but JP put in A LOT of work to let us use the full sweeps. This is what it is. I still recommend that you follow the initial script thread linked to on page 1 of the library thread (the original thread and not this one). There is a lot of good information there. The amount of thought behind and effort that went into the script is impressive.

 

[JP]

Great script!

However, I suggest adding the 0-20 Hz range to the measurement as well
There are so many different, generally stupid, guides on the Internet on how to match the cartridge compliance to the effective mass that people do it wrong, or do not do it at all.

It would be nice to see what the amplitude level of the unwanted resonance is, for example +1 dB or maybe +12 dB, (relative to the level of 1 kHz, on the graph)

Enough. It’s a phono cartridge response measuremnt tool, not an arm matching tool. Period.

 

[USER]

We do need a step by step tutorial in one post. Obviously also on the github page.

Now that you have it working perhaps you can help out and start a draft. We've been asking for support for 5 years now. I've been much too busy to update my instructions, which were for an older version of the script that worked in mono. I'm sure you realize how much effort this project has been for the few of us that have really taken it on. For me it was starting the library thread and working extra hard to make sure it did not die an early death or work on things such as the instructions. Not to mention mention measuring 50+ cartridges with extreme care and effort because if the early ones were not dependable then that too could kill the project.

I do hope that it is acknowledged that everyone that has put in a modicum of effort into and asked questions regarding the script has been helped and has been able to use it. The bitching is obnoxious.

 

[Quando]

Enough. It’s a phono cartridge response measuremnt tool, not an arm matching tool. Period.

Ok

 

[JP]

Regarding the chart generation script, it looks like this has been really well done and is in great hands. If there is ever a need for an extra hand there I have some undergraduate research experience doing data visualization and frequency-domain audio processing in Python, and several years of experience in backend Python development in general. Doesn't seem this is at all necessary, but seemed relevant and thought I should al least mention it.

From a data standpoint I want to move from single dimension lists to multi per dataset, and I think the chunk an concat routines are inelegant. I'm also looking at productizing custom visualizations, like plotting just the FR of multiple files.

Also considering (but not researched much) options for platform agnostic GUI to make it more accessible. First is to integrate the automatic processing of the audio from a single input file that is based on @DrCWO's work.

 

[undeakyetonde]

A suggestion: implement a constant lower limit for the distortion scale on the plot, something in the range of -60 or -70 dB

Currently, the lower limit seems to be automatically determined from data. This makes comparing measurements harder. In case of lower sample rates we get plots that (expectedly) extend down to -120 dB, which is not very useful.