"The Acusis is built around the XMOS XVF3000 voice processor, a single chip solution that supports up to four PDM microphones, USB and audio output, as well as on-board DSPs for beamforming and echo cancellation."
Acusis is a simple-to-use, complete solution for improving the audio quality of your speech-recognition or video communications project.
It solves two common issues, all in a single device that hooks up to your project with standard USB protocols.
conversational voice from across the room. 2. Echo cancellation: Acusis is designed to remove the "echos" of the sound your device is producing from what gets picked up by your microphone.
Also, Acusis is a linear array rather than circular, which makes it great for mounting on a monitor or TV.
If you’re building a project using a microphone today, or you just want better audio for your video chats or podcasts, Acusis is a great solution.
Acusis solves two really important problems for anyone building an audio/speech project:
Acusis is designed to capture normal speaking volume from across a room, due to its advanced beam-forming microphone array. The microphones form a "phased-array", working together to listen in the direction of the speaker. You may notice that the microphones are placed with varying spacing, that’s to help with beamforming. The spacing between a microphone pair is best for one frequency, so having different spacing helps cover more frequencies. Learn more about beamforming.
Acusis is designed to remove the "echos" of the sound your device is producing from what get picked up by the microphone, significantly reducing feedback in the case of voice/video conferencing, and keeping your speech recognizer from hearing the audio you are creating. All you have to do is make sure your audio output is passed through Acusis and it will make sure it doesn’t feed that audio back to you.
You attach Acusis to your computer/Raspberry Pi/custom system via USB. It will connect to your device as a USB microphone and a USB speaker. The reason it appears as a speaker is because to remove echos from the microphone input, it needs to receive the audio you are producing.
You can get audio to it two ways. One is to send audio to it as your primary audio output and connect your external speaker to Acusis’ line-level audio output. In this mode all of the sound is passed through to your speaker.
If you can’t put Acusis between your device and your speaker, you can usually configure your device to output audio to two devices at the same time, one output being Acusis, the other being the speakers you want to use.
On a Mac, you do this with "Audio MIDI Setup", and it’s pretty straight-forward. Other platforms have either built-in or third-party solutions (we’ll post solutions on our site).
We built Acusis to be attached to the front of a webcam; that’s why it has a circular opening in the center. Just use some double-sided tape and you’re good to go. At some point in time, we may introduce velcro attachments, stay tuned!
Acusis is built on a really cool component, the XMOS XVF3000. The XMOS is a single chip that lets us hook up four PDM microphones, USB, and audio output. Internally, it has an array of DSPs for running algorithms for beamforming and echo cancellation.
You can use Acusis out-of-the-box as a really good microphone. You can also hack on it yourself by downloading the SDK from XMOS and writing your own code. If you’re interested in this, add the XTAG programmer accessory to your pledge.
If we get enough interest, we will also build an on-line configurator to let you make minor adjustments to parameters without needing the SDK.
The audio processing is monophonic-only, which is fine for most applications of video conferencing or speech recognition. A stereo version may come later if this campaign is successful.
We designed the board to work without an enclosure to maximize the quality of microphone reception. There are mounting holes available if you decide to make an enclosure yourself.
Awhile back we built a voice-controlled application that ran on a Chrome browser in Linux and played podcasts. Our normal microphone picked up the podcast audio and fed it back into the speech recognition engine, which was pretty annoying since it filled up our logs with transcriptions of the podcasts. With Acusis, only our voices are heard, since the podcast audio output is cancelled by the XMOS DSP.
We’ve attached Acusis to our wall-mounted video chat system to make our audio communication better. We no longer need to yell at the microphone — Acusis hears quite well from across the room.
Bundle a Logitech camera and/or an XTAG programmer accessory with Acusis and start building your projects now.
|Re-Speaker Mic Array||Webcam Mic||Acusis|
|Automatic Echo Cancellation||No||No||Yes|
|Designed to Fit Over a Webcam||No||No||Yes|
We have already completed a prototype run using a manufacturer who can scale to a modest volume. We also have a contingency with a larger CM who can build high volumes to meet increased demand if needed.
Our team has been helping companies make products for more than three years, and most of us have 25+ years engineering experience working at places like Microsoft, Apple, Intel, Cisco, and Marvell. Even though this is the first time we’re running a Crowd Supply campaign, we understand the nuances involved in building and manufacturing products, and we have Crowd Supply to assist us along the way.
We have already designed, built, and tested Acusis, and have a few units ready right now. We’ve made sure we can get all the materials needed to build a volume order. Our primary "unknown" is how big the order is, which is why we are running this campaign. There are risks in that we may find certain parts have longer lead times to get in larger volumes, but we do have options to make component changes if that happens.
We’ll keep you posted throughout the process to give you the best estimate we can of when you’ll get your Acusis.
That said, our shipping dates are just estimates and it certainly could take us longer to ship than we hope. Also, Acusis works very well for us, but we can’t guarantee it will solve every audio problem you have, as your environment is probably different than ours. In other words: Acusis is offered on an as-is basis, and your mileage may vary. So don’t use Acusis in a commercial product without thorough testing, as we can’t be liable for how it works in your project.
We hope you enjoy using Acusis as much as we enjoyed making it!