Basic Functions and Model Selection¶
What is the process of developing products with Chipintelli’s ASR chip?¶
To begin development with Chipintelli’s ASR chips, follow these steps:
Initial Setup:
Choose the right chip/module from our documentation
For beginners: Purchase our evaluation kit
Development Phase:
Register on the Chipintelli AI Speech Development Platform
Develop firmware/software using our development guides
Test speech recognition and product functionality
Mass Production:
Option 1: Purchase pre-built modules in bulk
Option 2: Design custom PCBs using our hardware specifications
Testing & Validation:
Perform performance and reliability testing
Follow Document Center’s testing protocols
Ensure compliance with mass production standards
Support:
For assistance, contact us via:
- Platform’s online support
- Email: support@Chipintelli.com
- Direct technical support
How to select the appropriate speech chip or module when developing products?¶
You can click ☞Hardware Selection Guide to obtain information. If your product category is covered in our solution documentation, you can follow the recommended chip/module selections provided there for your development needs
The board shape provided by Chipintelli does not meet the requirements, can we provide customized design?¶
You have two options for custom board design:
- Self-Design Option: Purchase our ASR chips and design your own board using our hardware development documentation.
- Custom Board Service: For qualifying order quantities, we offer custom board design services. Contact our sales team for minimum order requirements and pricing.
Are there any recommendations for speech chips with 200 entries?¶
Our CI1103 ASR chip can support up to 300 voice commands. For the number of commands supported by different chips, please refer to☞Hardware Selection Guide.
How to select the number of microphones for speech products?¶
Our company has single microphone and dual microphone options, both of which support high recognition rate. The cost of single microphone is relatively low, while the cost of dual microphone is slightly higher. Dual microphones support sound source location, noise suppression and other functions. Single microphone is recommended for general applications; Dual microphones are recommended for applications in noisy environments and products with sound source positioning requirements.
How to improve speech recognition?¶
There are many factors that affect the effect of speech recognition, such as the performance of the ASR chip itself, the choice of commands, and the structural design of the product (or the installation position of the microphone and speaker). It is recommended that you directly use Chipintelli’s ASR chip products and AI speech development platform for development. We offer a wide range of technologies and products applications, and the recognition rate of the standard product solution can reach 97%. You can also directly use the platform’s optimization and other functions to improve the speech recognition effect.
How to support various dialect recognition?¶
We offer a wide range of Chinese dialects recognition solutions. For unknown dialects, we can also support recognition through self-learning algorithm. Please check the details☞offline self-learning function . Consult our technical team for more information.
What is the power consumption of Chipintelli chip? Does it support battery applications?¶
For instant response, our offline speech recognition requires the chip to remain in standby mode. The CI110X series (CI1102/CI1103) has the following power characteristics:
Standby power consumption: ~10mA
Active power consumption: 35-50mA
Startup time to recognition: ~0.5 seconds
Battery-Powered Solutions:
For battery-powered applications, we offer two modes:
- Always-On Mode: Use a high-capacity battery to maintain standby power
- Button-Activated Mode: Press button to wake the module and enable speech recognition; Turn off when not in use to conserve battery
This flexibility allows for efficient power management in various battery-operated applications.
What are the advantages of offline speech processing over online speech processing?¶
Our offline speech processing solution leverages dedicated ASR chips to handle voice commands directly on the device, eliminating the need for an internet connection. Key benefits include:
- Enhanced Privacy: All voice processing occurs locally, ensuring user data never leaves the device
- Network Independence: Provides reliable operation regardless of internet connectivity
- Rapid Response: Delivers consistent sub-1-second response times
- Wide Applicability: Ideal for smart home devices, automotive systems, lighting controls, and other IoT applications
This solution is particularly effective for voice-controlled functional products where reliability and speed are critical.
What are the main advantages of the hybrid offline-online voice solution compared to pure online or pure offline solutions?¶
The hybrid offline-online voice solution offers the best of both worlds by integrating local and cloud-based processing:
Key Benefits:
1. Reliable Core Functionality: Essential voice commands work offline, ensuring uninterrupted operation
2. Enhanced Security: Sensitive voice data remains on-device for basic commands
3. Cloud-Connected Features: Access to advanced AI services and content when online
4. Network Resilience: Maintains core functionality during internet outages
5. Rich User Experience: Combines the reliability of offline processing with the power of cloud services
Ideal For:
- Smart home ecosystems
- Voice assistants with content services
- Products requiring both reliability and advanced features
This solution is particularly valuable for devices that need to maintain core functionality while offering access to cloud-based services and content.
How to implement voice control linkage for smart home devices?¶
You can adopt a voice AIoT solution, which uses smart voice chips for voice control functionality while adding wireless modules like WiFi or Zigbee for device interconnection, enabling voice control of all connected smart home devices.
How to select WIFI for Chipintelli’s hybrid offline-online voice solution?¶
When selecting WIFI for the hybrid solution, consider the following:
- WIFI hardware interface should support UART (baud rate above 921600)/IIS SLAVE/standard SDIO DEVICE.
- WIFI CPU frequency should be above 160MHz, with more than 200KB SRAM allocated for the hybrid application layer.
- WIFI SDK should support basic protocols like JSON, SPEEX, Mbedtls, MQTT, HTTP, etc.
For software design, please refer to the relevant content in the SDK development section of the documentation center.
Which cloud services are available for Chipintelli’s hybrid intelligent voice solution?¶
Our company has integrated and tested the following cloud platforms for quick support:
- Tencent Cloud Xiaowei AI Voice
- QiWu Cloud AI Voice
- Turing Cloud AI Voice
- ROOBO Cloud AI Voice
These cloud platforms provide online AI voice services, including natural language interaction for music, weather, smart home, and other online content services. They support core services such as speech recognition (ASR), natural language processing (NLP), and text-to-speech (TTS). We also have partners supporting more cloud platforms. Please contact our business personnel for details.
Does the hybrid voice recognition solution support remote control (IoT) functionality?¶
Our current hybrid voice recognition solution can integrate IoT functionality and connect to IoT clouds (such as Tuya Cloud, Tencent Cloud, etc.) to enable mobile phone remote control of products. We also offer mature voice AIoT product solutions for users to choose from.
Can the device be awakened by using other than the wake word?¶
Intelligent voice products use wake words to maintain a balance between functionality and false recognition rate. Under normal circumstances, only the wake word can trigger the device. To accommodate different accents, the wake word system is designed with certain flexibility to ensure better recognition rates, so that similar-sounding words can occasionally trigger the device. For optimal accuracy and compatibility, further optimization of the wake word design is recommended.
Further our company offer wake-word self-learning technology, which enable users to train their own wake-word.
I purchased a recording board and a development board from your company. How should I connect them?¶
Please connect the IIS signal lines from the development board to the recording board. Use a mini-A cable or a USB-to-UART tool’s 5V output to power the development board, and use another mini-A cable to power the recording board, connecting both to the PC’s USB interface.
How does your company’s module generate voice firmware?¶
You can find relevant development instructions in the software development section of this documentation center and generate firmware through our AI speech development platform.
How can I customize the voice prompts for your voice module?¶
Voice prompts can be customized through our AI speech recognition platform, which also provides text-to-speech synthesis.
Your voice module’s microphone comes with wires. Can I directly mount the microphone on the board? Which method is better?¶
Soldering time and temperature can affect the microphone’s sensitivity. Therefore, we recommend using the wired connection method. If you choose surface mounting, please confirm with the microphone manufacturer whether it can withstand reflow soldering.
How to configure single/dual microphones in the B03GS01S module’s SDK?¶
This module only supports a single microphone. For dual-microphone solutions, please use the B03GT01S module.
How to connect the microphone and speaker to the B02GS01S module? Is there a polarity requirement?¶
For the microphone, the “+” terminal should be connected to the left channel of the module. Incorrect installation will significantly reduce sound levels and affect recognition rates.
For speakers:
- Single speaker: No polarity requirement, either connection will work.
- Dual speakers (left/right): Must be connected correctly. SPKL+ should connect to the positive terminal of the speaker to avoid phase cancellation and audio issues.
How to configure UART in the B03GS01S voice module’s SDK?¶
The commonly used communication UARTs are uart1/uart2. In user_config.h, there are two macros: UART2_ENABLE and UART2_USEUART1_ENABLE.
- To use uart2: Set UART2_ENABLE to 1
- To use uart1: Set both UART2_ENABLE and UART2_USEUART1_ENABLE to 1
Are there any principles to follow when setting up voice command terms?¶
Recommended guidelines for command terms:
- Length: 4-6 characters
- Avoid similar-sounding vowels
- Choose easily pronounced characters
- Minimize highly colloquial or frequently used words
What to do if the firmware flashing progress stays at 0% or reports an error?¶
Common solutions for these issues:
-
For CI110X/CI112X series chips:
- Pull up the PGEN pin to 3.3V
- Connect the serial tool to the computer
- Check the corresponding COM port in the upgrade tool
- Power cycle the device to start the download -
Check the signal level of the TX pin:
- CI110X/CI112X chips don’t support 5V or higher UART signals
- If there’s a level mismatch, modify the serial tool’s output level or add a level-matching circuit -
Check power supply stability:
- Some devices use 5V from the serial tool, which might cause insufficient power during startup
- Use a separate power supply for stable operation -
Check connection reliability:
- Loose or oxidized DuPont wire connections can cause issues
- Replace the connecting wires if necessary
How to choose the right USB-to-UART converter?¶
For reliable firmware flashing with Chipintelli’s offline voice chips, use a USB-to-UART converter that supports stable transmission rates up to 2Mbps. We recommend converters using stable communication chips like CH340G or FT232 to minimize flashing issues and improve efficiency.
How to resolve error warnings during firmware packaging?¶
If you encounter errors during firmware packaging:
1. Check the error message
2. For partition size errors (e.g., “User partition: File too large, exceeds reserved size”):
- Adjust the “Reserved Size” in the User options according to the actual size shown in the error message
- Increase in increments of 0x1000
- Enter the correct parameters and refresh the address to repackage the firmware
- For similar errors in other partitions, adjust their sizes accordingly
For more information, press F1 in the tool to view the help documentation.
What module is suitable for snoring detection?¶
You can use a module equipped with the CI1302 chip with SDK version 2.5.28. If training is required, please contact our sales team.
Which chip models support TTS functionality?¶
Please contact our sales team for this information.
Can it recognize dialects?¶
Yes, but training is required for dialects like Cantonese, etc.
Does CI1306 support speech-to-text applications?¶
No, it does not.
Does it support fuzzy command words?¶
Yes, it supports both word generalization and natural speech patterns.
Can CI1302 perform speech recognition for text other than command words?¶
No, it cannot.
Can the chip support both speech recognition and text-to-speech output simultaneously?¶
No, it cannot.