[data,fs,bit] = wavread('test.wav'); %//Read in WAV file
%//Note if you have dual channel or multi-channel, you need to do each channel separately, or add them to create mono-channel. In my example, which is voice, I use a single channel recording.
teststft = spectrogram(data); %//This is rough, and usually results in terrible results;
%//I suggest you set your input parameters suitable for voice/music/etc - Note I used hamming window of 256 length, 50% overlap and 512 nfft - which creates roughly 30ms time frames, suitable for voice.
teststft = spectrogram(data(:,1),hamming(256),50,512);
%//In your plotting example, you have used something like the Log Magnitude response, which can be expressed as:
testSTFT = log(abs(teststft));
%//and finally, to plot:
figure
surf(STFT, 'edgecolor', 'none'); view(0,90); axis tight;
xlabel('Time');
ylabel('FrequencyDistribution');
grid on;
1 回答
叹息......我没有给出Matlab
specgram和spectrogram的情节示例，但我会给出另一种方法 .首先，您要使用FFT将时域信号转换为短时频域（STFT）：
这可以通过matlab
specgram来完成：结果如下：
