หมายถึงงานนี้ใช้เครื่องวัดยี่ห้อ SPL จากตามในลิงค์ที่เอามาให้อ่านหรือครับ?
ผมขอถามคุณวิทยาหน่อยว่า ตอนทำ Waterfall Spectrum นั้นป้อนสัญญานแบบไหนครับ? พิงค์น๊อยส์หรือเปล่า? เพราะถ้าใช้วิธีสวีปความถี่ คงไม่ได้แกน z
ผมเคยแต่ทำวอเตอร์ฟอลล์เพื่อดูคุณสมบัติของเสียงจากการเปิดเพลงแค่นั้น ไม่เคยทำเพื่อเช็ครูมโมดแบบนี้ครับ
ท่าทางคุณPinIt จะไม่ได้อ่าน ผมเอามาให้อ่านที่นี่แล้วกันครับ
How REW makes its measurements
REW uses the logarithmic sine sweep measurement method. You can find out a great deal about the log sweep method, and the various alternatives, by reading the paper "Transfer Function Measurement with Sweeps" by SWEN MÜLLER and PAULO MASSARANI at
http://www.anselmgoertz.de/Page10383/Monkey_Forest_dt/Manual_dt/aes-swp-english.PDF, but I'll give a basic explanation here. If you'd rather not know, you can skip the explanation.
To make a measurement we need a sound source (a loudspeaker or subwoofer) and a microphone (SPL meters contain a microphone and can be used instead of a mic). A logarithmic sweep signal is sent to the source, which is a tone that starts at a low frequency and whose frequency increases steadily to a higher frequency. What makes the sweep logarithmic is the rate at which the frequency changes, it takes a fixed time to double (for example, the time for the sweep to go from 20 to 40Hz is the same as the time to go from 40 to 80Hz). The mic picks up the sweep, hearing the sound that travels to it directly from the source but also the sound that gets to it by bouncing off surfaces in the room first.
When we have captured the sound the mic picked up, the analysis starts. A process called "Fast Fourier Transform" (FFT) is used to calculate the individual frequencies (their amplitudes and phases) that made up the sweep we sent to the source (its spectrum). The same process calculates the amplitudes and phases of the frequencies in the signal the mic picked up. By comparing the amplitudes and phases of the signals the mic saw with those the sweep contained we can work out how each frequency has been affected by the room we are measuring. This is is called the "Transfer Function" of the room from the location of the source to the location of the mic - note that at a different source position or different mic position there will be a different transfer function, our measurement is only valid for one specific source and mic position. Having worked out the transfer function we can use an "inverse FFT" to get from the frequency amplitude and phase information to a time signal that describes the way any signal is changed when travelling from the source to the mic. That time signal is called the "impulse response" -
the transfer function it is derived from, it is only valid for one specific source and mic position.
The impulse response is actually exactly the same signal we would see if we could emit a very short but loud click at the source position and record what the mic picks up afterwards ("very short" meaning lasting just the time of 1 sample at the sample rate we are using for our analysis, so for a 48kHz sample rate that would be just 1/48,000 of a second which is 21 millionths of a second). You might ask why we don't just use a click then. One difficulty is that the click, because it is so short, needs to be extremely loud for us to be able to pick up what happens after the initial click over the background noise of the room. We could no longer use a speaker to generate that, we would need something
a starting pistol or to pop a balloon. We would also need a mic that could cope with both the extremely loud click itself and the much quieter echoes of the click produced by the room. You are likely to find that your family and neighbours are not that keen on you repeatedly firing a pistol to figure out what your room is doing, and even if they put up with it your results would not be as good as using a sweep. To be technical about it, you can achieve a much higher signal-to-noise (S/N) ratio by the sweep method. The S/N is determined by the background noise level and by how much energy is in the test signal, which in turn depends on how loud the signal is and how long it lasts. An impulse is extremely short, a few millionths of a second, so to get any signficant energy it needs to be very loud. A sweep can last for many seconds, so even at a modest volume its total energy can approach as much as a million times more than an impulse.
Once the impulse response has been obtained, it can be analysed to calculate information about how the room behaves. The simplest analysis is the FFT, to show the frequency response between the source and mic positions. However, we have some control over it. Altering which part of the impulse response is analysed by the FFT changes what aspect of the room's response we see. The early part of the impulse response corresponds to the direct sound from the source to the mic, the shortest path between them. Sound that has bounced off the room's surfaces has to travel further to reach the mic, which takes longer, so the later parts of the impulse response contain the contributions of the room. "Windowing" the impulse response to look at only the initial part shows us the frequency response of the direct sound with little or no contribution from the room. A window that includes later parts of the response lets us see how the room's contribution alters the frequency response. The ability to separate the contributions of the direct and later (reflected) sound is a key difference between the frequency response derived from an impulse response and one we would get from an RTA, for example, which can only show the total combined response of source and room.
Other information we can get from the impulse response include a "waterfall" plot, which is generated by moving a window in steps along the response and plotting the various frequency responses in order to produce a 3D picture of the way the response changes over time, and the room's "RT60" data, which is the time it takes sound in various frequency bands to decay by 60dB (meaning 1,000 times smaller than it was).
