spinorama.org

[delta76]

sure. I can easily add stuff with Claude (or Gemini).

Few suggestions out of my head:
- clicking on the photo should open the speaker's page to default spin
- there should be a link to category, for example if I go here https://www.spinorama.org/speakers/AsciLab C8C/ASR/index_asr.html there should be clickable links to Ascilab, bookshelf, and active speakers, so I go to those categories
- search could be improved, for example if I search for kef q meta https://www.spinorama.org/?page=1&search=kef+q+meta no result is returned.
- Spin page like this one https://www.spinorama.org/speakers/AsciLab C8C/ASR/index_asr.html should be centered, not align right

Again, those are my nice-to-have improvements. Keep up the good work!

 

[mianake]

Thanks very much for all your hard work. I am pretty clueless, and really appreciate these scores even if I don't understand them.

One question: I see there are multiple reviews for some speakers, and some variation. I would suspect this could be due to many factors, but I curious if there was something on this point. I might note my current thinking is getting Ascend, Ascilab or similar scored speakers for 1000-1500. Thank you very much.

 

[Schollaudio]

What does the tonality score mean? I'm sure it's explained somewhere but didn't see iy.

 

[pierre]

Thanks very much for all your hard work. I am pretty clueless, and really appreciate these scores even if I don't understand them.

One question: I see there are multiple reviews for some speakers, and some variation. I would suspect this could be due to many factors, but I curious if there was something on this point. I might note my current thinking is getting Ascend, Ascilab or similar scored speakers for 1000-1500. Thank you very much.

Each review shows some variation in measurements. The measurements are also very sensitive around placement and sometimes you see in the contour plot that the the microphone was not exactly aligned with the acoustic center. There are variations item to item that may be significative. New high quality brands have much tighter tolerance but most of the cheap speakers have item to item differences (in dB not tenth of a dB).

Currently the EQ is based on a smooth version of the default measurement. It could be valid to average all the high quality measurements together and optimise that.

 

[Matias]

What does the tonality score mean? I'm sure it's explained somewhere but didn't see iy.

Check below.

Spinorama.org – Spinorama

Help for the spinorama.org website. How to understand the graphs? What is a Spinorama? Why you should care?

www.spinorama.org

 

[pierre]

What does the tonality score mean? I'm sure it's explained somewhere but didn't see iy.

Tonality score is another name for Harmann/Olive score. The is a longer description in the help page.

Few suggestions out of my head:
- clicking on the photo should open the speaker's page to default spin

done

- there should be a link to category, for example if I go here https://www.spinorama.org/speakers/AsciLab C8C/ASR/index_asr.html there should be clickable links to Ascilab, bookshelf, and active speakers, so I go to those categories

done, you have a jump to menu on the right that allow exactly that

- search could be improved, for example if I search for kef q meta https://www.spinorama.org/?page=1&search=kef+q+meta no result is returned.

done

- Spin page like this one https://www.spinorama.org/speakers/AsciLab C8C/ASR/index_asr.html should be centered, not align right

done

see https://github.com/pierreaubert/spinorama/commit/5b7cda5d74ccefbc8abd37f7356fd6a6221817f7 for detailed changes (PR)

 

[delta76]

terrific, thanks for quick turn around!

 

[pierre]

I did a quick hack with some ml vision and other classical stuff.

I took data from @amirm, generated a picture. The tool then takes the picture and extract data to generate a new plot. The idea is to automate the ingestion of graphs for which we only have a picture like distorsion which have been asked for many many times.

for a first shoot: continuous lines are the original, dashed lines are extracted.

so ... with a bit of training i should get better results soon but it is alredy better than expected.

 

[pierre]

@pierre , only the "show annotations" box works on Firefox, neither the "show trend lines" or the "show recommended zones" tick boxes work. All three show by default, but it's impossible to remove trend lines or recommended zones.

confirmed. i will look into it.

 

[sigbergaudio]

Since you're in a steam of fixes, the double click to isolate trace doesn't seem to work either. For instance on the CEA2034 graph, it's impossible to isolate only the listening window (as an example), but it appears that it should be possible. But at least I'm not able to make it work (Chrome).

 

[pierre]

Since you're in a steam of fixes, the double click to isolate trace doesn't seem to work either. For instance on the CEA2034 graph, it's impossible to isolate only the listening window (as an example), but it appears that it should be possible. But at least I'm not able to make it work (Chrome).

it is working for me. Can you tell me which Chrome version on which OS?

 

[sigbergaudio]

it is working for me. Can you tell me which Chrome version on which OS?

Windows 11, Version 145.0.7632.161 (Official Build) (64-bit).

Am I somehow doing it wrong?

It says to double click on legend. If I click once, all traces disappear, if I double click, nothing happens.

 

[pierre]

Windows 11, Version 145.0.7632.161 (Official Build) (64-bit).

Am I somehow doing it wrong?

It says to double click on legend. If I click once, all traces disappear, if I double click, nothing happens.

View attachment 517278

ok found the issue. it works if the legend is below the graph but not when it is on the right...

 

[sigbergaudio]

ok found the issue. it works if the legend is below the graph but not when it is on the right...

Aha. Interesting bug. I tried making the window smaller, but the legend doesn't move below the graph still. What determines where it ends up?

 

[amarsicola]

Hi Pierre, first of all thank you very much for your work !
My suggestions are:
1) by chatting with Gemini, it suggests you could call a Gemini API to have prices updated (a good prompt would also fix the price per unit/pair issue)
2) the ranking could be complemented with a "value chart" dynamic scatter with price on X axis and preference score on Y axis, in which you could highlight a "best value pareto-frontier", this frontier consists of speakers where no other speaker provides a higher Preference Score for a lower Price. The chart could also change according to filters... I just sketched an example with Gemini

3) would love to filter ranking by -3db

Kudos !!!

 

[pierre]

Hi Pierre, first of all thank you very much for your work !
My suggestions are:
1) by chatting with Gemini, it suggests you could call a Gemini API to have prices updated (a good prompt would also fix the price per unit/pair issue)

the prb is to get it consistent, so it is a lot more than 1 call and i need a framework to catch regression.
I am trying and see what i will get. Gemini is horribly slow for me.

2) the ranking could be complemented with a "value chart" dynamic scatter with price on X axis and preference score on Y axis, in which you could highlight a "best value pareto-frontier", this frontier consists of speakers where no other speaker provides a higher Preference Score for a lower Price. The chart could also change according to filters... I just sketched an example with Gemini

View attachment 517427

good idea, there are some similar statistics in https://docs.google.com/spreadsheet...kJHh56iaJHm4/edit?gid=879425941#gid=879425941 but i can easily implement it. Having a proper set of statistics has been on the backburner for a long time.

In the stat page, you now have a new value chart which should be what you want. Filters will come soon in the custom tab.
Custom should allow people to plot whatever they want over time. Currently it is a bit buggy but can improve quickly.

Weight v.s. score

there is plenty of glitches i am aware of.

3) would love to filter ranking by -3db

done (and other parameters that were missing).

Kudos !!!

Thanks

 

[tifune]

Hi @pierre, the "similar" option doesn't work for the ascilab C8C, but it does for every other ascilab speaker. Is there something different in the C8C measurements causing this?

 

[sigbergaudio]

@pierre are we supposed to be able to click on the column headers on the ranking table to change sorting? Doesn't work for me.

 

[pierre]

@pierre are we supposed to be able to click on the column headers on the ranking table to change sorting? Doesn't work for me.

you can sort with the menu today but i am adding sort in the header too.

Hi @pierre, the "similar" option doesn't work for the ascilab C8C, but it does for every other ascilab speaker. Is there something different in the C8C measurements causing this?

that's a bug, now fixed. The closest is the JBL M2 (and very very close).

 

[sigbergaudio]

@pierre And the ranking doesn't appear to be complete? Or how does it work? The Sigberg speakers doesn't seem to be on it?