Loading module…กำลังโหลดโมดูล…
About this simulatorเกี่ยวกับโปรแกรมจำลองนี้
This convolution visualizer shows discrete-time convolution the way it is taught by hand: flip the kernel h, slide it across the input x, multiply the overlapping samples and add them up. Every step of y[n] = Σ x[k]·h[n−k] is drawn on the signal graph and written out as arithmetic.โปรแกรมแสดงภาพคอนโวลูชันนี้อธิบายคอนโวลูชันแบบไม่ต่อเนื่องแบบเดียวกับที่คำนวณด้วยมือ: พลิกเคอร์เนล h เลื่อนผ่านสัญญาณอินพุต x คูณแซมเปิลที่ซ้อนทับกัน แล้วบวกเข้าด้วยกัน ทุกขั้นของ y[n] = Σ x[k]·h[n−k] ถูกวาดบนกราฟสัญญาณและเขียนเป็นการคำนวณให้เห็น
What you can do hereสิ่งที่ทำได้ในหน้านี้
- Scrub or play the shift n and watch the flipped kernel slide under the input signal.เลื่อนหรือกดเล่นค่าการเลื่อน n แล้วดูเคอร์เนลที่พลิกแล้วเลื่อนอยู่ใต้สัญญาณอินพุต
- Drag any input sample to edit the signal and see the output update instantly.ลากแซมเปิลอินพุตใดก็ได้เพื่อแก้ไขสัญญาณ แล้วดูเอาต์พุตเปลี่ยนทันที
- Compare kernels: moving average, triangular, exponential decay and the difference [1, −1].เปรียบเทียบเคอร์เนล: ค่าเฉลี่ยเคลื่อนที่ สามเหลี่ยม เอกซ์โพเนนเชียลลดลง และผลต่าง [1, −1]
- Check the output length N + M − 1 and see which samples overlap at every shift.ตรวจสอบความยาวเอาต์พุต N + M − 1 และดูว่าแซมเปิลใดซ้อนทับกันในแต่ละการเลื่อน
Concepts coveredหัวข้อที่ครอบคลุม
- Discrete convolutionคอนโวลูชันแบบไม่ต่อเนื่อง
- Impulse responseผลตอบสนองอิมพัลส์
- Moving-average filterตัวกรองค่าเฉลี่ยเคลื่อนที่
- Commutativityสมบัติการสลับที่
- Output length N + M − 1ความยาวเอาต์พุต N + M − 1
Common questionsคำถามที่พบบ่อย
- How do you compute a convolution graphically?คำนวณคอนโวลูชันด้วยวิธีกราฟอย่างไร?
- Flip h[k] to get h[−k], shift it by n to get h[n−k], multiply it sample by sample with x[k], and sum the products. Repeat for every n. The simulator animates exactly these four steps.พลิก h[k] ให้เป็น h[−k] เลื่อนไป n ให้เป็น h[n−k] คูณกับ x[k] ทีละแซมเปิล แล้วรวมผลคูณ ทำซ้ำสำหรับทุกค่า n โปรแกรมนี้แสดงแอนิเมชันของสี่ขั้นตอนนี้
- Why is the output longer than the input?ทำไมเอาต์พุตจึงยาวกว่าอินพุต?
- The kernel starts overlapping the input before the first sample and stops after the last, so an N-sample input convolved with an M-sample kernel gives N + M − 1 output samples.เคอร์เนลเริ่มซ้อนทับอินพุตก่อนแซมเปิลแรกและสิ้นสุดหลังแซมเปิลสุดท้าย อินพุตยาว N แซมเปิลคอนโวลูชันกับเคอร์เนลยาว M แซมเปิลจึงได้เอาต์พุต N + M − 1 แซมเปิล