Millet AI smart speaker far-field voice interactive technology solutions

Millet AI speaker far-field voice interaction program not only includes 6-meter ring array technology and far-field wake-up technology, but also gives it a unique custom features such as Dual-wake, Free-cut, One-shot. Including beam-forming (beam-forming), noise suppression, reverberation suppression, array gain, echo cancellation and other technologies to help millet AI speaker effectively shield noise interference, to achieve long-distance precision man-machine conversation. The following is the detailed information of the millet AI speaker far-field speech interaction program.

(1) SAI_MICAK_60_3229_EVK (6-Mega-Low-Cost Solution)

SAI_MICAK_60_3229_EVK far-field voice interactive board development program, widely used in intelligent speaker, DOT, TV box and other voice intelligent products, using a highly integrated and cost-effective integrated design, with omnidirectional wake, source direction, directional pickup, Noise suppression, reverberation cancellation, echo cancellation, far-field speech recognition, semantic understanding, and other technologies, support HDMI 2.0 ([email protected]), DLNA, Miracast wireless projection, embedded Ula intelligent systems and massive content services, support Alexa, DuerOS, Mor, and other third-party intelligence systems support ODMs and OEMs to help customers quickly implement AI product development.

Basic parameters:

• 6 wheat ring array, support broadcast interruption, arbitrary interruption, continuous wake-up

• Sound source positioning: 360° (horizontal direction), positioning accuracy: ±10° (horizontal direction)

• Signal-to-noise ratio> 65dB, noise suppression> 20dB, sensitivity> -42dBV @94dB 1kHz

• 96% wake-up distance, 5 meters awakening rate> 91%

• 95% recognition distance, 5 meters recognition rate> 90%

• Stereo low-power amplifier, infrared gesture recognition, three-color ring light effect, WI-FI/BT+BLE, 3.5MM Jack, USB/UART, etc.

(2) SAI_MICAK_40_102_EVK (4 wheat low cost solution)

SAI_MICAK_40_102_EVK far-field voice interaction suite development solution, widely used in intelligent speaker, TV box and other voice intelligent products, using a microphone array board, main control board and power amplifier board split design, with omnidirectional wake, sound source direction, directional pickup Audio, noise suppression, reverberation cancellation, echo cancellation, far-field speech recognition, semantic understanding and other technologies, embedded Ula intelligent systems and massive content services, support Alexa, DuerOS, Mor third-party intelligent systems, support ODM and OEM To help customers quickly realize AI product development.

• 4 wheat ring array, support broadcast interruption, arbitrary interruption, continuous wake-up

• Sound source positioning: 360° (horizontal direction), positioning accuracy: ±15° (horizontal direction)

• Signal-to-noise ratio> 65dB, noise suppression> 20dB, sensitivity> -42dBV @94dB 1kHz

• 96% wake-up distance, 4m awakening rate> 91%

• 95% recognition distance, 4 meters recognition rate> 90%

• Stereo high-performance amplifier, monochrome ring effect, WI-FI/BT+BLE, UART, etc.

(3) SAI_MICAK_60_16_EVK (6-Mega High Performance Solution)

SAI_MICAK_60_16_EVK far field voice interaction kit development solution, widely used in intelligent speaker, TV box and other voice intelligent products, using a microphone array board, main control board and power amplifier board split design, with omnidirectional wake, sound source direction, directional pickup Audio, noise suppression, reverberation cancellation, echo cancellation, far-field speech recognition, semantic understanding and other technologies, embedded Ula intelligent systems and massive content services, support Alexa, DuerOS, Mor third-party intelligent systems, support ODM and OEM To help customers quickly realize AI product development.

• 6 wheat ring array, support broadcast interruption, arbitrary interruption, continuous wake-up

• Sound source positioning: 360° (horizontal direction), positioning accuracy: ±10° (horizontal direction)

• Signal-to-noise ratio> 65dB, noise suppression> 20dB, sensitivity> -42dBV @94dB 1kHz

• 96% wake-up distance, 5 meters awakening rate> 91%

• 95% recognition distance, 5 meters recognition rate> 90%

• Stereo Low Power Amplifier, Monochrome Ring Effect, WI-FI/BT+BLE, UART, etc.

Awake word + speech semantic recognition integrated method to achieve a seamless connection between awakening words and voice manipulation. One sentence easily realizes all the skills of the millet AI speaker and becomes the real life assistant of the user.

related suggestion:

Recommended: What Smart Speakers Do Smart Speakers Use Inventory

A: Install WeChat fingerprint lock Beijing password lock sales Hubei smart lock to join