Fundamentals of Product Development¶
The development of the electronic product scheme mainly has the difference between hardware and software. The hardware scheme generally refers to the circuit board designed for the product using chips and other electronic components. The software scheme generally refers to the code developed based on the type of the main control chip on the circuit board and the functional requirements of the product, and then the firmware that can be downloaded to the main control chip is generated by compiling tools, and the specific functions of the product are realized by cooperating with the hardware work such as chips. If the user wants to develop a new product scheme, the main steps include hardware selection, hardware development, firmware development, and product scheme testing. The user can complete the development after repairing the software and hardware problems found in the test, and then carry out the subsequent production and manufacturing according to the requirements.
1. Basic concept¶
This chapter describes some terms and concepts involved in the product development scheme using Huawei chips, as follows:
1.1 ASR chip¶
Chips are also called integrated circuits, which are called integrated circuits in English, or IC for short. A chip is a device that can be directly welded by miniaturizing circuits and some electronic components, manufacturing them on the surface of semiconductor wafers through special equipment, and then processing them later. The use of chips in the hardware scheme can greatly reduce the area of the circuit board and the complexity of the design, help to speed up product development and reduce the difficulty of development, and can save the cost of the overall scheme. Therefore, a large number of chips have been applied to almost all electronic product fields such as household appliances, automobiles, consumer electronics, industrial control, etc.
ASR chip is a kind of chip, which provides corresponding computing power through the neural network technology at the chip end. It can directly process speech related functions such as speech recognition at the terminal without network server. Intelligent speech products directly use intelligent speech chips, which can greatly improve the intelligent speech experience, improve the development efficiency, and reduce the overall cost of the scheme. Therefore, more and more intelligent devices are using professional intelligent speech chips to develop products.
1.2 Speech recognition process of ASR chip¶
Generally speaking, when the ASR chip needs to process speech recognition, it first needs the peripheral microphone to collect speech signals, convert the acoustic signals into electrical signals, and then send the sampled digital signals to the speech processing module inside the ASR chip through ADC sampling on the ASR chip. At the same time, the chip executes the corresponding recognition control and algorithm software to calculate the model of the sound, After decoding by software, the recognition result is output in text mode.
Users can click ☞Learn about speech recognition principles to obtain more relevant knowledge information.
1.3 Speech module¶
The speech module is to design the intelligent speech chip and the required passive components on a small PCB board, which can be connected to the microphone speaker and other peripheral devices. The module can download the corresponding functional firmware to form a hardware circuit board with basic speech functions.
If the designed speech module is directly used in the product development and design, it is unnecessary to learn and care about the hardware knowledge required for designing the circuit of the intelligent speech chip, nor to design the speech circuit part of the product, to avoid the hardware board problems caused by improper processing of speech related signals, and to accelerate the hardware development of the product. Therefore, for new users, It is recommended to directly use our ready-made standard modules for product scheme design.
1.4 Firmware¶
Firmware generally refers to the binary files that can be directly downloaded to the main chip through the compiler and other tools after the user has developed the functional software. The binary files are mainly machine code instructions that can be recognized by the main chip. After the main chip is powered on, it will execute the internal instructions of the firmware to achieve the software functions that the user wants to achieve.
In the scheme development of our intelligent speech chip, the firmware corresponding to our chip is mainly composed of software control part, acoustic model, language model, broadcast feedback, etc. The software control part includes the instructions of the operating system and each peripheral of the control chip to execute the corresponding functions, including task scheduling, serial communication, control IO input and output, etc; The acoustic model and language model are documents required for speech recognition, which can be obtained by users through our AI development platform; Broadcast feedback refers to the sound that the scheme needs to play after executing a speech command to inform the user that the command has been executed.
In actual product development, since our company’s intelligent speech chip may be used as the master chip or controlled chip, the firmware can be divided into master firmware and slave firmware, as follows:
- Main control firmware: As the main control of the product, the intelligent speech chip can control the function of the product according to the recognized speech commands while conducting speech recognition and recognition feedback.
- Slave firmware: as the controlled chip, the intelligent speech chip communicates with the upper computer (product master chip) (such as serial communication), uploads the speech recognition results to the upper computer, and can accept the instructions of the upper computer for broadcast feedback.
2. Hardware selection¶
The hardware selection includes chip selection or module selection. If the user plans to design speech related circuits by himself, the hardware selection is mainly to select appropriate chips; If the user only designs the peripheral circuit and the speech related circuit directly uses the module, the hardware selection is mainly to select the appropriate module. The following is an example of our products, which are described as follows:
2.1 Select chip¶
The user can sort out the corresponding hardware specifications of the product (such as how many microphones, how much memory space, how much control IO, etc. are required) according to the requirements of the required application scenarios (such as product type, number of entries, number of announcements, noise environment applications, whether speech orientation is supported, whether end NLP is supported, whether master chips are required, etc.), Refer to the specific specifications of our chips in ☞Hardware Selection Guide, and select appropriate chips for design.
If the user does not know how to choose, we recommend that you check the relevant documents in ☞Product Scheme Development to obtain more information for you to choose the right chip. If you are still unclear, you can contact us through the following contact information for help.
Tel.: +86-028-61375925 Email: support@chipintelli.com
2.2 Selecting Modules¶
As mentioned in Section 2.1, after sorting out the actual application scenario requirements and hardware specifications, users can refer to the specific specifications of our modules in the ☞Hardware Selection Guide to select appropriate modules for design.
If you don’t know how to choose, you can also check the relevant documents in ☞Product Scheme Development to obtain more information for your selection, or contact us through the following contact information for help.
Tel.: +86-028-61375925 Email: support@chipintelli.com
2.3 Sample acquisition methods of chips and modules¶
Our company provides a channel to purchase chips and module samples. After selecting the appropriate chips or modules, users can click ☞sample purchase to obtain samples of chips or modules.
If the user is a new user in this field, it is recommended to directly use our third-generation speech chip, such as CI1302 or CI-D02GS02S module, for product development. Our company also provides a development board kit, CI-D06GT01D, which is convenient for development. Users can also click the above purchase link to purchase
3. Hardware development¶
After the user completes the hardware selection, he/she can choose from our company ☞Chipintelli Speech AI Development Platform Get the hardware reference documents, including schematic diagrams and PCBs, and refer to the design methods in the documents for hardware board level design.
The user can also view the ☞Hardware Design Reference document to obtain the corresponding design knowledge, so as to reduce the possibility of self design errors, improve the development effect and speed up the development efficiency.
4. Firmware development¶
Our company has provided software SDK development kit and corresponding tools based on intelligent speech chips. Please refer to the Software Development SDK section of the Document Center for details, which will not be repeated here. When developing firmware, new users should first register our speech AI platform account ☞click to enter Chipintelli Speech AI Development Platform, and then obtain the SDK and tools corresponding to the selected chip or module from the platform. We recommend that new users first use our development boards or standard modules, directly compile the SDK provided by our company, use tools to produce firmware, and complete the development process. ☞Click to view the example of CI130X using platform to make firmware and download. The document center contains the specifications of the modules selected by the user. The specifications include the hardware connection and handling methods for module upgrading. The user can operate according to the documents.
For new users, we recommend that users directly use the firmware production function of the Chipintelli Speech AI Development Platform to directly generate downloadable firmware. This can avoid the use of IDE tools for code development and is helpful to get started quickly. Please click the ☞Chipintelli Speech AI Development Platform User’s Guide document to use the corresponding operation method in this document.
If users who are familiar with our development need to separately synthesize various components of the firmware, then package and generate the firmware, and need to do some code development at the same time, they can download the corresponding SDK according to the selected chip for development. For SDK development instructions, refer to:
5. Product testing and feedback¶
After the firmware developed by the user is successfully burned on the hardware board, the user can power on the hardware and say the wake-up word or command word to confirm and test its recognition effect. For the test method, refer to the relevant documents in ☞Product Test.
If the user finds a bad command word during the test, he can compress the bad audio (zip/rar) and send an email to the support@chipintelli.com And other ways to feed back to us, and we will quickly give a reply after analysis.
We look forward to using our products to design excellent product solutions. We will provide you with comprehensive technical support to help your products gain more advantages in the market