Specialized in the technology industry, LAPIS speech synthesis LSI makes the car “knowing”

What will the car of the future look like? Autonomous driving, Internet of Vehicles, safety, and comfort all put forward higher and higher requirements for the intelligentization of automobiles.

Voice, as an important way of human-computer interaction, will undoubtedly play an increasingly important rigid-needed role in scenes such as cars that require the driver’s eyes to focus ON the front.

To realize the voice interaction between people and vehicles, speech synthesis solutions will become an indispensable part. The latest in-vehicle speech synthesis LSI “ML2253x series” launched by LAPIS, a subsidiary of Rohm Group, is specially designed for in-vehicle speech synthesis. The response time to 5ms and the error self-check function make the vehicle voice function truly safe and reliable. Whether it is ADAS (Advanced Driver Assistance System) or AVAS (Vehicle Approach Alerting System), the “ML2253x series” can be used to great effect.

Specialized in the technology industry, LAPIS speech synthesis LSI makes the car “knowing”

Lapis Stone – Lucky Stone

LAPIS semiconductor Co., Ltd. started Semiconductor manufacturing as OKI Electric Industry Co., Ltd. in 1961, separated and reorganized in 2008, became a member of the ROHM Group as OKI Semiconductor, and changed its name to LAPIS Semiconductor in 2011. The name “LAPIS” comes from the Latin word LAPIS for a gemstone “lapis lazuli”, which means overcoming challenges and bringing luck. LAPIS Semiconductor mainly focuses on low-power technology, digital-analog mixed-load technology, high-frequency circuit technology, memory design technology, high-Voltage process technology, ultra-small packaging technology, etc. LAPIS Semiconductor has 40 years of profound accumulation in the field of speech synthesis LSIs. The global annual sales of speech synthesis LSIs are about 15 million pieces, ranking first in the Japanese market share.

Vehicle Voice Output: MCU Solution vs Speech Synthesis LSI

There are two main ways to realize the playback of in-vehicle voice: one is realized through MCU, and the other is realized through speech synthesis LSI. Realized by MCU, speech synthesis is mainly realized by MCU+software, and additional memory and Amplifier are required; while LAPIS’s speech synthesis LSI solution, speech playback is realized by hardware, and the chip integrates memory, amplifier, logic operation and I/F, etc. By directly connecting the speaker and MCU, the MCU only needs to issue control commands without participating in speech synthesis, which greatly reduces the burden on the MCU.

Specialized in the technology industry, LAPIS speech synthesis LSI makes the car “knowing”

Ms. She Mengyi from ROHM Semiconductor (Shanghai) Co., Ltd. introduced that the market for in-vehicle voice output systems is now increasing with the implementation of laws and regulations related to automatic emergency braking systems (AEBS) and the types of ADAS warnings used in the market. The demand for in-vehicle voice has also increased. On the basis of meeting conventional vehicle regulations, the performance of the voice output system is critical in two aspects: response time and fault detection.

When the voice output system is realized by MCU, the software design is very complicated and there are many parts. When the main control MCU performs processing other than voice playback, it may not be able to perform voice broadcast, and the problem of voice broadcast delay may also occur. The speech synthesis LSI solution launched by LAPIS Semiconductor solves these problems: (1) The speech synthesis LSI is composed of hardware, does not load the main control MCU, and can play the voice within 5 ms after the command is sent; (2) ) Built-in components required for voice playback, which can reduce the number of components in the system as a whole; (3) Improve the failsafe (Fail Safe) function.

What’s special about “ML2253x series”

The newly launched “ML2253x series” chips of Lapis Semiconductor include two types, one is the M22Q53x series with built-in Flash, and the built-in Flash is available in 2MB, 4MB, and 16MB; the other is an external memory. The ML22530 series can be expanded up to 128MB.

Ms. She Mengyi introduced that the “ML2253x series” has four new features:

(1) Playing sound abnormality detection function: The new products of “ML2253x series” are newly equipped with the playing sound abnormality detection function. This function will re-convert the analog signal output by the speaker into a digital signal, and compare it with the playback data inside the LSI. When an abnormal situation higher than the set value is detected, an error signal will be fed back to the main control MCU. It is very helpful to improve the quality of the voice output system.

(2) The function of rewriting the Flash memory by using the main control MCU: On the basis of the previous on-board rewriting function using the ROM writer, the rewriting function through the main control MCU has been added. The main control MCU can be used to write data into the built-in Flash memory via SPI I/F, not only can the voice data be rewritten during the development process, but also the voice data can be rewritten after the product leaves the factory. Using this function, it can also support the needs of a new generation of vehicles that will be closely connected to the Internet in the future and gradually realize IoT.

(3) HQ-ADPCM compression method: There is a contradictory trade-off between high sound quality and memory saving, and the new product adopts HQ-ADPCM compression method, which takes into account and realizes these two advantages at the same time. In the past, only PCM and ADPCM2 compression methods were used, and sound effects and music would be seriously degraded during compression, or they could only be uncompressed. The new product adopts HQ-ADPCM method, even if the memory capacity is compressed to 1/5 of the previous compression method, it can play a smooth sound normally. The HQ-ADPCM compression method can be selected through the Speech LSI Utility software in the Sound Device Control Suite.

(4) Various functions required for in-vehicle LSIs: The new product supports high-temperature operation (105°C) required for in-vehicle use and AEC-Q100, the reliability standard for automotive Electronic products. In addition, it is equipped with abnormal detection functions such as short circuit detection, high temperature abnormality detection, and command transmission error detection that were installed in previous models. These functions also greatly contribute to improving the reliability of the in-vehicle voice output system.

There are specialties in the technology industry. The launch of the “ML2253x series” makes the realization of vehicle voice output faster and more reliable, and the “speaking” cars are gradually approaching people.