How does Sphere compare to the classic microphones we modeled? That’s perhaps the question most asked and which, of course, deserves an answer.
Central to the development of Sphere were countless listening tests and comparisons, with the various vintage mics that we modeled. Over the last few weeks, in particular, we hosted several listening sessions with producers and engineers. Here are some comments:
"As a lover and owner of lots of vintage mics, I was extremely skeptical but intrigued to try this mic out. The C12 emulation was quite frankly mind-blowing and worth many times more than the cost of this mic alone. Sold!"
Dave Izumi - Ed Harcourt, Duke Special, James Bay, Toploader.
“The C12 emulation was stunning. Even on its own basic setting it was really impressive for the projected prices..... Plus the spatial sphere tweaks are really interesting...."
Paul Godfrey, Morcheeba
Now You Can Take The Test
Today we are releasing the Townsend Labs Sphere L22 A/B comparison tests, which directly compare the Sphere microphone models and the original microphones the models are based on. Each set of audio files comes in an "A" version and a "B" version, where the model and the original are randomly assigned to A or B, so that the test is blind.
Comparison files are available for download in 48kHz 24-bit WAV format. In your DAW, load the “A” files on one track and the corresponding “B” files in another track. Then you can solo either track to compare them (In Pro Tools use XOR solo mode).
We realize no comparison test is definitive, but hopefully, this will provide a good indication of Sphere's modeling accuracy. We plan on doing more comparison tests with a wider range of source material and use cases.
How The Audio Was Recorded
These comparisons were recorded using Universal Audio Apollo (silverface) preamps and converters with Unison processing disabled. In the future, we also plan on doing more tests which include various other vintage and modern preamps.
The vocal test files compare a mid-1950s era U47 with the Sphere model at both cardioid and omni polar pattern settings, as well as a closer distance option for the cardioid setting. This particular 47 is one of those rare specimens that the first owner put in a closet for decades. The second, and current, owner was lucky enough to realize what they had and purchased it at a bargain price.
For this test, the U47 and the Sphere L22 were placed right next to each and carefully aligned for the most equal sound pickup. Inevitably there are some errors in the tests due to it being impossible for the microphones to occupy the same point in space, but we tried to minimize as much as we could.
This 47 is not the same mic that the Sphere LD-47K model is based. Even though this 47 sounds great and is in great condition, we found another 47 that we thought sounded just a bit better, but that one was not available at the time of the listening test.
We also did comparison tests using a guitar amp. These comparisons include an M49, a U47 (a different one than that used on the vocal files), a U67, and a U87. Where applicable we have omni, cardioid and figure-8 patterns, with both on-axis and 45 degrees off-axis positions. The M49, U67 and U87 models in the Sphere plug-in are based on the specific microphones you are hearing in the guitar amp listening test.
For this test, one microphone was recorded at a time and the guitar amp was re-amped so that exactly the same sound is reproduced for every recording. All microphones were carefully aligned using a self-leveling alignment lasers, so the mic capsules are in exactly the same position for every test. This is one of the most accurate ways to do a comparison test because the sound source is highly repeatable and the microphones can be positioned in exactly the same location.
What's the Difference?
We also put together difference files which subtract the B file from the A file, so you can literally hear the difference between the model and the original. If the models are perfect then you should just be left with a little bit of reverb. For the vocal comparisons where the mics weren't in exactly the same location, this is especially true.
The vocal comparisons are also interesting because in a few cases you can hear when the vocalist moves slightly and the files don't cancel as well. It goes to show how sensitive these types of tests can be to mic position. In a couple of the guitar files, you can also hear a pop or two where the Twin Reverb tube amp was acting up.
Which Is Which?
The answers to which files contain the original and which contain the model will be released at a later date. Thanks for taking part!
Chris Townsend, CEO and CTO
Townsend Labs Inc.