NEW
Products
Featured Products
Product Lines
Deliver reliable products for acoustic measurement and testing
Sensors
Provides measurement microphones, mouth simulators, ear simulators, and more for accurate acoustic measurements.
Data Acquisition
Combines hardware and software for high-speed, high-precision signal acquisition, ideal for various acoustic applications.
Acoustic Imaging
Offers acoustic cameras for gas leak detection, partial discharge, and fault diagnostics across handheld, fixed, and UAV platforms.
Noise Measurement
Includes sound level meters, noise sensors, and monitoring systems for effective noise measurement and analysis.
Electroacoustic Test
Delivers complete electroacoustic testing solutions, including analyzers, testing software, and acoustic test boxes.
Solutions
Provide high-quality solutions for the acoustic field
Blogs
Share insights, cases and trends in acoustic testing
During pilot production and production line ramp-up, many issues do not appear in the way teams initially expect. Sometimes it starts with a small fluctuation at a test station, or a comment from a line engineer saying, "This result looks a bit unusual."However, when takt time, yield targets, and delivery milestones are all under pressure, these seemingly minor anomalies can quickly be amplified and begin to affect the overall production rhythm. We have been working with Huaqin as a long-term partner. As projects progressed, the challenges encountered on the production line became increasingly complex. On site, our role gradually extended from basic production test support to problem analysis and cross-team coordination during pilot production. In many cases, the focus was not simply on whether a test station was functioning, but on how to absorb uncertainties early and prevent them from disrupting delivery schedules. The following two experiences both took place during the pilot production phase of Huaqin projects. They are not exceptional cases. On the contrary, they represent the kind of everyday issues that most accurately reflect the realities of production line delivery. Airtightness Testing Issues in Project α During the pilot ramp-up of Project α, the airtightness test station for the audio microphone showed clear instability. For the same batch of products, pass rates fluctuated noticeably across repeated tests, frequently interrupting the station's operating rhythm. Initial troubleshooting naturally focused on the test system itself, including software logic, equipment status, and basic parameter settings. It soon became clear, however, that the issue did not originate from these areas. As on-site verification continued, we gradually confirmed that the anomaly was more closely related to the product's mechanical structure and material characteristics. This model used a relatively uncommon combination of materials. A sealing solution that had worked well in previous projects could not maintain consistency during actual compression. Even slight variations in applied pressure were enough to influence test results. Once the direction of the problem was clarified, the on-site approach shifted accordingly. Rather than repeatedly adjusting the existing solution, we returned to verifying the compatibility between materials and structure. Over the following period, we worked together with the customer's engineering team on the production line, testing multiple material options. This included different types of silicone and cushioning materials, variations in silicone hardness, and adjustments to plug compression methods. Each step was evaluated based on real test results before moving forward. The process was not fast, nor was it particularly clever. In essence, it came down to repeatedly confirming one question: could this solution run stably under real production line conditions?Ultimately, by introducing a customized soft silicone gasket and making fine parameter adjustments, the airtightness test results gradually stabilized. The station was able to run continuously, and the pilot production rhythm was restored. Figure 1. Test Fixture Diagram Noise Floor Issues in Project β Compared with the airtightness issue in Project α, the noise floor anomaly encountered during pilot production in Project β was more complex to diagnose. During headphone pilot production for Project β at Huaqin's Nanchang site, the noise floor test station repeatedly triggered alarms. Test data showed that measured noise levels consistently exceeded specification limits, significantly impacting the pilot production schedule. This model used high-sensitivity drivers along with a new circuit design, making the potential noise sources inherently more complex. It was not a problem that could be resolved by simply adjusting a single parameter. Rather than focusing solely on the test station, we worked with the customer's audio team to investigate the issue from a system-level signal chain perspective. The process involved sequentially testing different shielding cables, adjusting grounding strategies, evaluating various Bluetooth dongle connection methods, and isolating potential power supply and electromagnetic interference sources within the test environment. Through continuous spectrum analysis and comparative testing, the scope of the issue was gradually narrowed. It was ultimately confirmed that the elevated noise floor was primarily related to power interference from the Bluetooth dongle, combined with differences in product behavior across operating states. After this conclusion was reached, relevant configurations were adjusted and validated on site. As a result, noise floor measurements returned to a stable and controllable range, allowing pilot production to proceed. Figure 2. Work with the customer engineer to solve problems Common Characteristics of Pilot Production Issues Looking back at these two pilot production experiences, it becomes clear that despite their different manifestations, the underlying diagnostic processes were quite similar. Whether dealing with airtightness instability or excessive noise, the root cause could not be isolated to a single module. Effective resolution required on-site evaluation across mechanical structure, materials, system operating states, and test conditions. During pilot production, issues of this nature rarely come with ready-made answers. They are also unlikely to be resolved through a single verification cycle. More often, progress is made through repeated trials, comparisons, and eliminations, gradually converging on a solution that is genuinely suitable for long-term production line operation. Production line delivery rarely follows a perfectly smooth path. In many cases, what ultimately determines whether a project can move forward as planned are those unexpected issues that must be addressed immediately when they arise. In our long-term collaboration with customers, our work often takes place at these critical moments—working alongside engineering teams to stabilize processes, maintain momentum, and keep projects moving forward step by step. If you also want CRYSOUND to support your production line, you can fill out the Get in Touch form below.
In our previous blog post, "Abnormal Noise Detection: From Human Ears to AI"we discussed the key pain points of manual listening, introduced CRYSOUND's AI-based abnormal-noise testing solution, outlined the training approach at a high level, and showed how the system can be deployed on a TWS production line. In this post, we take the next step: we'll dive deeper into the analysis principles behind CRYSOUND's AI abnormal-noise algorithm, share practical test setups and real-world performance, and wrap up with a complete configuration checklist you can use to plan or validate your own deployment. Challenges Of Detecting Anomalies With Conventional Algorithms In real factories, true defects are both rare and highly diverse, which makes it difficult to collect a comprehensive library of abnormal sound patterns for supervised training. Even well-tuned—sometimes highly customized—rule-based algorithms rarely cover every abnormal signature. New defect modes, subtle variations, and shifting production conditions can fall outside predefined thresholds or feature templates, leading to missed detections (escapes). In the figure below, we compare two wav files that we generated manually. Figure 1: OK Wav Figure 2: NG Wav You can see that conventional checks—frequency response, THD, and a typical rub & buzz (R&B) algorithm—can hardly detect the injected low-level noise defect; the overall curve difference is only ~0.1 dB. In a simple FFT comparison, the two wav files do show some discrepancy, but in real production conditions the defect energy may be even lower, making it very likely to fall below fixed thresholds and slip through. By contrast, in the time–frequency representation , the abnormal signature is clearly visible, because it appears as a structured pattern over time rather than a small change in a single averaged curve. Figure 3: Analysis results Principle Of AI Abnormal Noise Algorithm CRYSOUND proposes an abnormal-noise detection approach built on a deep-learning framework that identifies defects by reconstructing the spectrogram and measuring what cannot be well reconstructed. This breaks through key limitations of traditional rule-based methods and, at the principle level, enables broader and more systematic defect coverage—especially for subtle, diverse, and previously unseen abnormal signatures. The figure below illustrates the core workflow behind our training and inference pipeline. Figure 4: Algorithm Flow Principle During model training, we build the algorithm following the workflow below. Figure 5: Algorithm Judgment Principle How To Use And Deploy The AI Algorithm Preparation First, prepare a Low-Noise Measurement Microphone / Low-noise Ear Simulator and a Microphone Power Supply to ensure you can capture subtle abnormal signatures while providing stable power to the mic. Figure 6: Low-Noise Measurement Microphone Next, you'll need a sound card to record the signal and upload the data to the host PC. Figure 7: Data Acquisition System Third, use a fixture or positioning jig to hold the product so that placement is repeatable and every recording is taken under consistent conditions. Finally, ensure a quiet and stable acoustic environment: in a lab, an anechoic chamber is ideal; on a production line, a sound-insulation box is typically used to control ambient noise and keep measurements consistent. Figure 8: Anechoic Room Figure 9: Anechoic Chamber Model Development First, create a test sequence in SonoLab, select "Deep Learning" and apply the setting. Next, select the appropriate AI abnormal-noise algorithm module and its corresponding API Figure 10: Sequence Interface 1 Then open Settings and specify the model type, as well as the file paths for the training dataset and test dataset. Click Train and wait for the model to finish training (Training time depends on your PC's hardware) Figure 11: Sequence Interface 2 During training, the status indicator turns yellow. Once training is complete, it switches to green and shows a "Training completed" message. Figure 12: Sequence Interface 3 Finally, place your test WAV files in the specified test folder and run the sequence. The model will start automatically and output the analysis results. Test Case Figure 13:Test Environment Figure 14:Test Curve System Block Diagram Figure 15: System Block Diagram 1 Figure 16: System Block Diagram 2 Equipment More technical details are available upon request—please use the "Get in touch" form below. Our team can share recommended settings and an on-site workflow tailored to your production conditions.
As A²B microphones and sensors are increasingly adopted in automotive applications, the demand for reliable testing in both R&D and production is also growing. This article explains why A²B testing matters, highlights the advantages of A²B over traditional analog cabling in terms of interconnect and scalability, outlines key measurement KPIs (such as frequency response, THD+N, phase/polarity, and SNR), and presents a typical test-bench setup along with the corresponding solution configuration. Why A²B Microphone and Sensor Testing Matters In-cabin audio is no longer just "music playback". Modern vehicles depend on high-performance acoustic sensing for hands-free calling, in-cabin communication, voice assistants, ANC/RNC, and more—and these features increasingly rely on multiple microphones and even accelerometers deployed around the cabin. ADI notes that the rapid expansion of audio-, voice-, and acoustics-related applications is a key trend, and that new digital microphone and connectivity approaches are enabling broader adoption. To deliver consistent performance, teams need a test workflow that is repeatable across different node positions, harness lengths, and configurations—without turning every debug session into a custom project. The Interconnect Shift: From Shielded Analog Cables to Digital A²B Historically, scaling microphone counts often meant scaling shielded analog cabling, which adds weight, cost, and integration burden—sometimes limiting these features to premium vehicle segments. A²B (Automotive Audio Bus) addresses that interconnect problem by enabling a scalable, networked digital audio architecture with deterministic behavior—exactly what timing-sensitive acoustic applications need. Figures a and b show how such a design may be realized with the traditional analog and the digital A²B systems, respectively. Figure 1 (a) Analog system design with analog mic elements (shielded wires). (b) Digital system design with digital mic elements (A²B technology and UTP wires). What You'll Measure: Key A²B Microphone KPIs Frequency Response (FR) THD+N Phase / polarity (and channel-to-channel consistency for arrays) SNR AOP (if required by your program/spec) Typical Block Diagram-What the Bench Looks Like At CRYSOUND, we provide more than just the CRY580 A²B interface. We offer a full automotive audio testing solution, including audio acquisition cards, microphones and sensors, acoustic sources, custom fixtures, acoustic test boxes, and vibration shakers, delivering a complete and streamlined testing experience. Figure 2 Here's a description of the testing block diagram, including the use of the latest OpenTest Audio Test & Measurement Software https://opentest.com Solution BOM List The value of end-to-end delivery: reducing system integration time and minimizing coordination costs between multiple suppliers. We cover everything from R&D to production line testing. Figure 3 BOM list of the solution If you'd like to learn more about A²B testing, please fill out the Get in touch form below and we'll reach out shoutly.
Precision measurement is only as trustworthy as the tools behind it. At CRYSOUND, long-term reliability has always been part of how we support professional acoustic testing and metrology work—especially for equipment expected to perform consistently over many years. That's why CRYSOUND provides a 10-Year Limited Warranty for eligibleCRY3000 series sensors offering long-horizon confidence for labs, manufacturers, and audio professionals who depend on stable performance. What the 10-year warranty covers This is a limited warranty focused on defects in materials or workmanship that occur under normal use, installation, and maintenance conditions. It is not a guarantee of fitness for a specific purpose. Eligible product categories (CRY3000 Series) The 10-year limited warranty applies to the following CRY3000 Series categories (traceable by product nameplate/serial number):Microphones Preamplifiers Microphone Sets Mouth Simulators Ear Simulators Ear Simulator Sets. Warranty term: 10 years (and what's different) For the main product categories above, the warranty term is 10 years.Accessories/consumables (e.g., windscreens, cables, adapters, seals, replaceable pinnae, packaging) are covered under a 6-month warranty unless otherwise specified by contract or separate terms. When the warranty period starts The warranty period is calculated from the shipping/delivery date. If that date is unavailable, it is calculated from the end-user purchase date (with contract proof). If valid proof cannot be provided, CRYSOUND may use the factory date or the latest traceable serial-number record as the basis. What CRYSOUND will do for eligible defects If CRYSOUND confirms the issue is covered, we may provide one or more of the following: Free repair, including necessary parts and labor Replacement with the same model, or a model of equal or higher performance (new or certified refurbished/remanufactured) Customized/project products follow contract terms Repairs or replacements do not extend the original warranty period. Clear boundaries (typical exclusions) As a limited warranty, it excludes issues caused by misuse, drops/crushing, liquid ingress, corrosive environments, out-of-spec power/ESD/surge, improper installation/grounding/sealing/maintenance, unauthorized disassembly/modification, missing/altered serial numbers, normal wear/cosmetic changes, shipping/storage mishandling, or third-party compatibility problems (where applicable). Calibration note (important for metrology users) Because microphones and simulators are metrology instruments, slight drift can occur due to environment and measurement uncertainty. Unless drift is confirmed to be caused by a manufacturing defect, calibration/recalibration and certificate updates are typically not included for free (paid calibration/verification services may be available). Service logistics (shipping & service location) For in-warranty cases, users typically cover round-trip shipping to CRYSOUND/authorized service points. Cross-border service may involve duties or customs clearance fees unless otherwise agreed by contract. CRYSOUND will arrange the nearest service option based on region, product type, and spare-part availability. Warranty & Support To request warranty service or technical support, contact info@crysound.com (or reach out to your CRYSOUND sales contact). See the warranty policy on our website: https://www.crysound.com/warranty/
