Command your lights, doors and home appliances Create voice-commanded remotes for TVs and media centers Reduce complexity and cost of control interfaces Speaker-dependent system keeps only you in charge
The SpeakUp is a speech-recognition board that either can plug into a mikroBUS socket on a host microcontroller system or be used standalone, connecting directly to external devices. You can set it up to recognize over 200 different voice commands and have the on-board MCU carry them out. It's speaker dependent, so it prefers to take orders only from you.
SpeakUp learns from scratch, so you can make it understand your commands spoken in any language, dialect or slang. Connect the SpeakUp to your PC via its mini-USB connector, then, using a simple software application, record a word or phrase (up to five seconds) and assign it an instruction. Repeat for as many commands as you need. When you put it to work, SpeakUp listens to what you're saying, matches the sound to one of the pre-recorded commands, and triggers the specified action which the on-board MCU executes.
What gives the SpeakUp its speech recognition capabilities is the firmware loaded in the on-board STM32F415RG MCU. The firmware is based on the DTW algorithm, which makes it decisive; it turns your talk into action almost instantly.
Sound is received either through the SpeakUp's on-board microphone or through an external microphone connected to board's 3.5mm jack. Between the mic and the MCU sits a VS1053 IC with a built-in stereo audio codec to process the raw signal.
Two output options After the processed sound has been forwarded to the STM32F415RG MCU which interprets the voice command, there are two output options that can be utilized at the same time or separately:
Header pins for both sets of connection pads are included which you can solder yourself.
- Click Mode sends the index of the matched voice command via the mikroBUS headers to a host system's USB or UART interface. As well as to mikroElektronika's development systems, you can connect SpeakUp to a PC, an Arduino, a Raspberry Pi, a BeagleBone Black, a Discovery, or a LaunchPad by using various adapter boards. See the mikroBUS Category description for links to the adapters.
- Standalone Mode allows the SpeakUp's STM32 MCU to control external devices directly via 12 user-programmable I/O headers.
Speech Recognition Board Software Recording commands and assigning actions is simple thanks to the software tool for programming the SpeakUp. First, connect the SpeakUp to your computer using a USB cable. It will show up as an HID device, so no special drivers are required. Just launch the app and start tinkering.
You start by creating a new project, which will hold all of your settings. Each voice command is shown as a tab on the software interface. You can add as many as you need. Recording for each command is performed right in the SpeakUp software. You can play back and re-record commands as needed, and name them whatever makes sense to you.
In the image to the right, you can see four commands have been recorded: ACTIVATE, OPEN, CLOSE and START. The ACTIVATE command's setup is showing. A description has been written for reference, and actions have been assigned to certain I/O pins. Note how you can assign aliases to the I/O pins (left column), as well as specify the type of action (middle column) and the duration/frequency of each action (right column).
The pin aliases are set in a General Settings area where you also can set the following:
- Deviation tolerance — Define how closely your delivery has to match your pre-recorded commands (five levels of adjustment).
- Buffer size — Set up the default length of your voice commands in a way that will use the SpeakUp's memory most efficiently.
- Threshold — Adjust the noise-gate sensitivity of the microphone to calibrate it for the noise level of the environment it will be used in.
- Notify master — Set how the commands will be issued to a host MCU if SpeakUp is connected to a host system via the mikroBUS headers.
- Data rate — Set the speed of the above notifications.
Speech Recognition Board Overview
Total thickness is 10.30 mm (405.50 mils).
- STM32F415RG ARM Cortex-M4 microcontroller with 1024kB Flash, 192kB RAM, 168 MHz operation
- Microphone with -44 ±3dB sensitivity, 100 Hz to 20 kHz frequency range, 56 dBa SnR
- VS1053 stereo audio codec with streaming support; line-out pads on board
- 12 programmable GPIOs, mikroBUS connector pads, and mini-USB connector for programming
For future-proofing, pads are on the board in which you can install a JTAG header compatible with the mikroProg for STM32, so the SpeakUp module's firmware can be updated if necessary.
Speech Recognition Board Resources
SpeakUp click Videos
Optional Recommended Products for this Item