{"id":69,"date":"2019-12-11T20:08:00","date_gmt":"2019-12-11T20:08:00","guid":{"rendered":"https:\/\/imitone.com\/blog\/?p=69"},"modified":"2024-05-03T18:05:27","modified_gmt":"2024-05-03T18:05:27","slug":"cause-for-calibration","status":"publish","type":"post","link":"https:\/\/imitone.com\/blog\/index.php\/2019\/12\/11\/cause-for-calibration\/","title":{"rendered":"cause for calibration"},"content":{"rendered":"\n

This post was transcribed from imitone’s kickstarter page. click here for the original post<\/a>, or click here for an audio version of this post.<\/a><\/em><\/p>\n\n\n\n

I’m calibrating!  I’m calibrating!!  I’m fine-tuning a training model, generating and analyzing billions of datapoints, calculating probabilities and building a time-critical pitch tracker capable of \u2014 ah \u2014 ah, wait, I shouldn’t get ahead of myself.<\/p>\n\n\n\n

I’ve been busy.  Following the plan from our last update, I have devoted most of this year to research and development on the next generation of imitone’s technology \u2014 probably the most advanced pitch tracker in the world.  A few weeks from now, it will be in your hands as part of imitone beta 0.10.<\/p>\n\n\n\n

Most exciting of all, it will be evolving<\/em>.<\/p>\n\n\n\n

\"\"\/
The new imitone technology uses statistics and machine learning to follow your voice’s pitch more precisely, especially when the sound is noisy or unclear.<\/figcaption><\/figure>\n\n\n\n

As imitone’s new “brain” develops, it will respond quicker, cleaner and more accurately to a wider range of voices and singing styles.  It will work in noisier spaces.  It will consume less CPU power \u2014 a big consideration for our mobile apps.  Finally, it will lay the foundation for important features like automatic scale detection, automatic pitch range and the intelligent note selection described in our last update.<\/p>\n\n\n\n

It has taken a long time and a lot<\/em> of work, but this is a huge step toward the dream.<\/p>\n\n\n\n

The coming update will bring a few of those benefits, and will be followed by several smaller updates with gradual improvements.  Next week, we’ll show you a preview of the new version and our plans for 2020.<\/p>\n\n\n\n

For now, read on to learn about what we’ve been doing in 2019, and how we’re building the new imitone.<\/p>\n\n\n\n

Evan’s 2019<\/h1>\n\n\n\n

Over the last year I’ve traveled across the USA in a motor home while continuing my research.<\/p>\n\n\n\n

January<\/strong>:  Finished the first version of our custom machine learning tool, the “calibrator”.  Showed imitone at NAMM 2019 with the MIDI Manufacturers Association.<\/p>\n\n\n\n

February<\/strong>:  Contributed to pitch-related features in MIDI 2.0.  Made plans for 2019 research, development and marketing.  Shot tutorial videos<\/a> with a new team member.<\/p>\n\n\n\n

March<\/strong>:  Contributed to MIDI 2.0 Property Exchange.  Designed a secret<\/em> planned feature.  Made another tutorial.  Developed marketing plans.  Attended GDC.<\/p>\n\n\n\n

April<\/strong>:  Lived in the Mojave desert.  Resumed work on the calibrator, training synthesizer, theory of pitch tracking and the ability to run imitone’s machine learning on cloud computers.  Made social media plans.<\/p>\n\n\n\n

\"\"\/
Mount San Jacinto, California, seen from Joshua Tree state park. In late April I took a spill in this landscape and sustained a concussion. I continued research, but shelved most side-projects for a while including my work with MIDI 2.0. I’m told that the best thing to do after a head injury is to avoid screens and intense focus for a few weeks… Whoops!<\/figcaption><\/figure>\n\n\n\n

May<\/strong>:  Lived on solar power in the mountains of California.  Tested several experimental SAHIR designs.  Developed mathematical techniques for high-fidelity filtering.<\/p>\n\n\n\n

June<\/strong>:  Visited with computational creativity experts in Santa Cruz.  Integrated the new filtering methods into imitone’s technology.  Used attractor theory<\/a> to improve imitone’s learning system.<\/p>\n\n\n\n

July<\/strong>:  Investigated and fixed several design flaws in the new SAHIR.  Started running imitone’s machine learning on cloud computers.  Visited Crater Lake in Oregon.<\/p>\n\n\n\n

August<\/strong>:  Further improved the training tools and the ability to run them in the cloud.  Made a significant improvement to the SAHIR’s pitch-following.<\/p>\n\n\n\n

\"\"\/
Crater Lake, Oregon, with Wizard Island in the background. Yeah, Wizard Island.<\/figcaption><\/figure>\n\n\n\n

September<\/strong>:  Developed a CPU budgeting algorithm for SoundSelf, which might be used in imitone too.  Interviewed potential additions to the imitone team.<\/p>\n\n\n\n

October<\/strong>:  Had an early holiday with family and caught the flu.  Planned out the 0.10 update.  Investigated the new SAHIR’s CPU usage.  Made a few “recipe” videos showing imitone techniques.<\/p>\n\n\n\n

November<\/strong>:  Made a pipeline for using the training data in the live SAHIR.  Began testing and tuning the new tech for real-world use.  Worked on 0.10’s interface design.  Developed a new approach to vibrato tracking and pitch-guide.<\/p>\n\n\n\n

December<\/strong>:  Getting the new technology ready for action and putting together imitone 0.10!<\/p>\n\n\n\n

Taking imitone to the Next Level<\/h1>\n\n\n\n

My focus with Phase 4 has been on unlocking the potential of imitone’s pitch tracking.  To do this, I’m adding a layer of machine learning to imitone’s original technology, and building a collection of tools to help me oversee the process.<\/p>\n\n\n\n

The Fundamentals<\/h1>\n\n\n\n

Let’s start with a quick summary of how imitone has always worked.<\/p>\n\n\n\n

If you sing beneath a concrete tunnel, into a plastic pipe or onto the strings of a guitar, these things will “resonate” with the tone of your voice.  If you sing at just the right pitch, the resonance will be loud, bright and long-lasting.  Otherwise, it will be quiet and dull.<\/p>\n\n\n\n

\n