news

New "LyraT-Mini" To Be Launched Soon

Shanghai, China
May 7, 2019

A new development board that is small in size, yet highly versatile and powerful, will soon be added to Espressif’s Lyra series.

Espressif has developed an ESP32-based audio development board that will be part of the Lyra series, which already includes ESP32-LyraT and ESP32-LyraTD-MSC. The new audio development board will be called LyraT-Mini, due to being very small in size and having a single-mic wake-up interrupt. Moreover, LyraT-Mini will have a simple hardware architecture and a very competitive price, coupled with an excellent keyword recognition rate.

LyraT-Mini consists mainly of the ESP32-WROVER-B module and a codec, as shown in the block diagram below.

ESP32-WROVER-B collects audio data from the microphone and achieves keyword recognition through such processes as noise cancellation (NC), automatic gain control (AGC), acoustic echo cancellation (AEC), etc., which are appropriately applied to voice-controlled IoT devices in different use cases.

LyraT-Mini Use Cases

The wake-word engine (WWE) has a high success rate even within a 5-meter distance between the user and the device. What's more, in a noisy environment (e.g., music volume: 85 db), the success rate of the wake-up interrupt can be as high as 81% in a distance of three meters between the user and the device. For more details about the success rates of LyraT-Mini, see the diagram below.

LyraT-Mini Success Rate

The default wake-up phrase in Chinese is “嗨乐鑫”,which translates as “Hi Espressif!” (pinyin: “Hi Lexin” / phonetics: “hʌɪ  ləʌˈʃɪən). However, LyraT-Mini offers customizable wake-up words, too, so customers can set up the ones that match their own preferences. The wake-up phrase can be used even when LyraT-Mini is in the offline mode, in order to turn on the power supply in the voice-controlled devices already connected to the board.

Not only does LyraT-Mini optimize the performance of the original functions of our existing ESP32-LyraT, but it also surpasses the need for voice processing at the pre-DSP stage. Furthermore, audio signal processing, such as acoustic echo cancellation (AEC), acoustic gain control (AGC), noise cancellation (NC), wake-word engine (WWE), etc., can be implemented on the ESP32-D0WD chip.

ESP-ADF is Espressif’s official audio development framework, on which LyraT-Mini is based. You can get all the details about ESP-ADF by clicking here. If you have any sales-related questions regarding LyraT-Mini, please submit them here.