listen carefully!

With an example, this page illustrates the difficulty to detect chirps which is the problem addressed in the paper. We give here several sound samples of a random chirp in random noise with the same set-up used in the benchmarking test of Sect. VI.C (p 16).

• Here is the noise free chirp .
• Here is the same chirp embedded in white Gaussian noise at SNR=50 .
• Same chirp in noise at SNR=20 .

Now, the idea is to compare our algorithm (the BCC, Best Chirplet Chain) to the human ears (well, only yours for now!).

Let's play! You will find below two sound samples. One is noise only and the other contains the chirp in noise with SNR=12. At this SNR, the BCC performs an almost sure detection for a false alarm of 1 into 10 (see ROC curve in Fig. 9 p 17). What about you? Listen and select the sample with the chirp!

sample 1
sample 2

[Note that the game is unfair since you could already ear the chirp in advance, so you have a reference while BCC does not!]