Skip to content

sgnl.sinks.impulse

A sink element that performs an impulse respose test.

ImpulseSink dataclass

Bases: TSSink

A sink element that performs an impulse response test

Parameters:

Name Type Description Default
original_templates Optional[str]

str, an h5 file name containing the templates before SVD

 None
template_duration Optional[float]

float, the duration of the templates, in seconds. This will inform us of how much data around the impulse to perform the impulse response test with.

 None
plotname Optional[str]

str, output file name prefix of plots

 None
impulse_pad Optional[str]

str, the sink pad name of the impulse data stream. This is used for plotting the impulse position.

 None
data_pad Optional[str]

str, the sink pad name of the filtered data stream to perform the impulse response with. When there are multiple data streams, only one data stream undergoes the impulse response test.

 None
verbose bool

bool, default False, be verbose

 False
bankno Optional[int]

int, the bank order number to choose which bank to perform the impulse test on. This order number is the one after the bank sorting. Many banks can be filtered but only one bank can under go the impulse response test.

 None
Source code in sgnl/sinks/impulse.py 
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
@dataclass
class ImpulseSink(TSSink):
    """A sink element that performs an impulse response test

    Args:
        original_templates:
            str, an h5 file name containing the templates before SVD
        template_duration:
            float, the duration of the templates, in seconds. This will inform us of
            how much data around the impulse to perform the impulse response test with.
        plotname:
            str, output file name prefix of plots
        impulse_pad:
            str, the sink pad name of the impulse data stream. This is used for plotting
            the impulse position.
        data_pad:
            str, the sink pad name of the filtered data stream to perform the impulse
            response with. When there are multiple data streams, only one data stream
            undergoes the impulse response test.
        verbose:
            bool, default False, be verbose
        bankno:
            int, the bank order number to choose which bank to perform the impulse test
            on. This order number is the one after the bank sorting. Many banks can be
            filtered but only one bank can under go the impulse response test.
    """

    original_templates: Optional[str] = None
    template_duration: Optional[float] = None
    plotname: Optional[str] = None
    impulse_pad: Optional[str] = None
    data_pad: Optional[str] = None
    verbose: bool = False
    bankno: Optional[int] = None

    def __post_init__(self):
        super().__post_init__()
        self.cnt = {p: 0 for p in self.sink_pads}
        self.A = Audioadapter(backend=TorchBackend)
        self.Ainput = Audioadapter()

    def pull(self, pad, bufs):
        """
        getting the buffer on the pad just modifies the name to show this final
        graph point and the prints it to prove it all works.
        """
        super().pull(pad, bufs)
        # bufs = self.preparedframes[pad]
        # FIXME: use preparedframes
        self.cnt[pad] += 1
        impulse_offset = bufs.metadata["impulse_offset"]
        self.impulse_offset = impulse_offset
        # if bufs.EOS:
        #    self.mark_eos(pad)
        padname = pad.name.split(":")[-1]
        if padname == self.data_pad:
            if bufs.buffers is not None:
                if self.verbose:
                    print(self.cnt[pad], bufs)
            for buf in bufs:
                if buf.end_offset > impulse_offset:
                    if buf.offset < impulse_offset + Offset.fromsec(
                        self.template_duration + 1
                    ):
                        # only save data around the impulse
                        buf.data = buf.data[self.bankno]
                        self.A.push(buf)
                    else:
                        recovered_impulse_offset, match = self.impulse_test()
                        print(f"{impulse_offset=} {recovered_impulse_offset=} {match=}")
                        if impulse_offset == recovered_impulse_offset and match > 0.997:
                            print("Impulse test passed")
                        else:
                            print("Impulse test failed")
                        self.mark_eos(pad)
                else:
                    # not at the impulse yet
                    pass
        elif padname == self.impulse_pad:
            for buf in bufs:
                if buf.offset > impulse_offset - Offset.fromsec(
                    1
                ) and buf.offset < impulse_offset + Offset.fromsec(
                    self.template_duration
                ):
                    # only save data around the impulse
                    self.Ainput.push(buf)

    def impulse_test(self):
        """
        filter output against time reversed template
        """
        # TODO: find a way to determine bankno
        # subbanks = bank['subbanks']
        # print("srates", subbanks[bankno]["rates"])
        # if n == -1:
        #    n = subbanks[bankno]["ntemp"]
        # ntot = subbanks[bankno]["ntemp"]
        # print(f"Running impulse test for {n} templates out of {ntot} total...")
        # self.full_template_length = 2048 * (1+4+8)

        f1 = h5py.File(self.original_templates, "r")
        full_templates = np.array(f1["full_templates1"])
        f1.close()

        nfull_temp = full_templates.shape[0]
        # assert nfull_temp == subbanks[bankno]["ntemp"], (
        #    "Template number does not match, check the --bankno given"
        #    + " to make sure the same bank is being compared"
        # )
        # print(nfull_temp, ntot)
        n = nfull_temp
        if self.verbose:
            print(
                "number of templates",
                nfull_temp,
            )
        filter_output = self.A.copy_samples(self.A.size)
        if self.verbose:
            print("filter_output shape", filter_output.shape)
        # bankno=0
        # filter_output = filter_output[: nfull_temp // 2].cpu().numpy()[bankno]
        # filter_output = filter_output.cpu().numpy()[bankno]
        filter_output = filter_output.cpu().numpy()

        # only filter with current number of time slices
        # full_template_length =  2048 * 13
        # full_templates = full_templates[:, -full_template_length :]

        if self.verbose:
            print("full_templates.shape", full_templates.shape)
        outid = 0
        full_templates_flipped = np.flip(full_templates, 1)

        response = []

        cmaximumnormeds = np.zeros(shape=int(n / 2))
        istart = int((nfull_temp - n) / 4)
        iend = istart + int(n / 2)
        imiddle = int((istart + iend) / 2)
        # complex
        if self.verbose:
            print("Calculating response")
        for i, k in enumerate(range(istart, iend)):
            # template pairs
            real = full_templates_flipped[2 * k]
            imag = full_templates_flipped[2 * k + 1]

            h = np.array(real + imag * 1j)
            normh = np.linalg.norm(h)

            realo = filter_output[2 * k]
            imago = filter_output[2 * k + 1]
            o = np.array(realo + imago * 1j)
            normo = np.linalg.norm(o)

            # correlate
            response1 = np.abs(correlate(o, h, "valid")) / normh / normo
            response.append(response1)

            # find peak
            cmaximum = response1.max()
            cmaximumnormeds[i] = cmaximum
            if k == imiddle:
                outid = np.where(response1 == cmaximum)[0][0]

        cavgn = np.average(cmaximumnormeds)

        plotname = self.plotname
        if plotname is not None:
            # for plotting response
            if self.verbose:
                print("Plotting...")
            m = imiddle
            im = int(n / 4)
            # output = np.pad(filter_output[m].real, (self.impulse_position, 0),
            # "constant")
            output = filter_output[m].real
            # res = np.pad(response[im], (self.impulse_position, 0), "constant")
            res = response[im]
            # indata = np.pad(
            #    torch.cat((self.indata)).cpu(), (self.impulse_position, 0), "constant"
            # )
            indata = self.Ainput.copy_samples(self.Ainput.size)
            # indata = np.zeros(2048)
            data = [indata, output, res] + [
                full_templates[m],
                full_templates_flipped[m],
            ]
            names = ["input", "output", "response"] + [
                "full\ntemplate",
                "full\ntemplate\nreversed",
            ]
            self.plot_wave(data, names, plotname, cmaximumnormeds, cavgn)

        return Offset.fromsamples(outid, 2048) + self.A.offset, cavgn

    # Plotting
    def plot_wave(self, data, dataname, figname, matchdata, cavgn):
        """
        Plot the input and output of the impulse response
        """
        plt.figure(figsize=(18, 10))
        n = len(data)
        maxlen = max(data, key=len)
        maxlen = len(maxlen)
        for i in range(n):
            plt.subplot(n, 1, i + 1)
            plt.plot(data[i])
            plt.xlim(0, maxlen)
            plt.ylabel(dataname[i])
            plt.tick_params(bottom=False, labelbottom=False)
        plt.tick_params(bottom=True, labelbottom=True)
        plt.tight_layout()
        plt.savefig(figname + "response")
        plt.clf()

        # Plot the match across templates
        plt.figure(figsize=(6, 4))
        plt.plot(matchdata, ".")
        plt.title("impulse response")
        plt.xlabel("template id")
        plt.ylabel("match")
        plt.axhline(cavgn, color="red", label=f"avg:\n{cavgn}")
        plt.legend()
        plt.tight_layout()
        plt.savefig(figname + "match")
        plt.clf()

impulse_test()

filter output against time reversed template

Source code in sgnl/sinks/impulse.py 
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
def impulse_test(self):
    """
    filter output against time reversed template
    """
    # TODO: find a way to determine bankno
    # subbanks = bank['subbanks']
    # print("srates", subbanks[bankno]["rates"])
    # if n == -1:
    #    n = subbanks[bankno]["ntemp"]
    # ntot = subbanks[bankno]["ntemp"]
    # print(f"Running impulse test for {n} templates out of {ntot} total...")
    # self.full_template_length = 2048 * (1+4+8)

    f1 = h5py.File(self.original_templates, "r")
    full_templates = np.array(f1["full_templates1"])
    f1.close()

    nfull_temp = full_templates.shape[0]
    # assert nfull_temp == subbanks[bankno]["ntemp"], (
    #    "Template number does not match, check the --bankno given"
    #    + " to make sure the same bank is being compared"
    # )
    # print(nfull_temp, ntot)
    n = nfull_temp
    if self.verbose:
        print(
            "number of templates",
            nfull_temp,
        )
    filter_output = self.A.copy_samples(self.A.size)
    if self.verbose:
        print("filter_output shape", filter_output.shape)
    # bankno=0
    # filter_output = filter_output[: nfull_temp // 2].cpu().numpy()[bankno]
    # filter_output = filter_output.cpu().numpy()[bankno]
    filter_output = filter_output.cpu().numpy()

    # only filter with current number of time slices
    # full_template_length =  2048 * 13
    # full_templates = full_templates[:, -full_template_length :]

    if self.verbose:
        print("full_templates.shape", full_templates.shape)
    outid = 0
    full_templates_flipped = np.flip(full_templates, 1)

    response = []

    cmaximumnormeds = np.zeros(shape=int(n / 2))
    istart = int((nfull_temp - n) / 4)
    iend = istart + int(n / 2)
    imiddle = int((istart + iend) / 2)
    # complex
    if self.verbose:
        print("Calculating response")
    for i, k in enumerate(range(istart, iend)):
        # template pairs
        real = full_templates_flipped[2 * k]
        imag = full_templates_flipped[2 * k + 1]

        h = np.array(real + imag * 1j)
        normh = np.linalg.norm(h)

        realo = filter_output[2 * k]
        imago = filter_output[2 * k + 1]
        o = np.array(realo + imago * 1j)
        normo = np.linalg.norm(o)

        # correlate
        response1 = np.abs(correlate(o, h, "valid")) / normh / normo
        response.append(response1)

        # find peak
        cmaximum = response1.max()
        cmaximumnormeds[i] = cmaximum
        if k == imiddle:
            outid = np.where(response1 == cmaximum)[0][0]

    cavgn = np.average(cmaximumnormeds)

    plotname = self.plotname
    if plotname is not None:
        # for plotting response
        if self.verbose:
            print("Plotting...")
        m = imiddle
        im = int(n / 4)
        # output = np.pad(filter_output[m].real, (self.impulse_position, 0),
        # "constant")
        output = filter_output[m].real
        # res = np.pad(response[im], (self.impulse_position, 0), "constant")
        res = response[im]
        # indata = np.pad(
        #    torch.cat((self.indata)).cpu(), (self.impulse_position, 0), "constant"
        # )
        indata = self.Ainput.copy_samples(self.Ainput.size)
        # indata = np.zeros(2048)
        data = [indata, output, res] + [
            full_templates[m],
            full_templates_flipped[m],
        ]
        names = ["input", "output", "response"] + [
            "full\ntemplate",
            "full\ntemplate\nreversed",
        ]
        self.plot_wave(data, names, plotname, cmaximumnormeds, cavgn)

    return Offset.fromsamples(outid, 2048) + self.A.offset, cavgn

plot_wave(data, dataname, figname, matchdata, cavgn)

Plot the input and output of the impulse response

Source code in sgnl/sinks/impulse.py 
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
def plot_wave(self, data, dataname, figname, matchdata, cavgn):
    """
    Plot the input and output of the impulse response
    """
    plt.figure(figsize=(18, 10))
    n = len(data)
    maxlen = max(data, key=len)
    maxlen = len(maxlen)
    for i in range(n):
        plt.subplot(n, 1, i + 1)
        plt.plot(data[i])
        plt.xlim(0, maxlen)
        plt.ylabel(dataname[i])
        plt.tick_params(bottom=False, labelbottom=False)
    plt.tick_params(bottom=True, labelbottom=True)
    plt.tight_layout()
    plt.savefig(figname + "response")
    plt.clf()

    # Plot the match across templates
    plt.figure(figsize=(6, 4))
    plt.plot(matchdata, ".")
    plt.title("impulse response")
    plt.xlabel("template id")
    plt.ylabel("match")
    plt.axhline(cavgn, color="red", label=f"avg:\n{cavgn}")
    plt.legend()
    plt.tight_layout()
    plt.savefig(figname + "match")
    plt.clf()

pull(pad, bufs)

getting the buffer on the pad just modifies the name to show this final graph point and the prints it to prove it all works.

Source code in sgnl/sinks/impulse.py 
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
def pull(self, pad, bufs):
    """
    getting the buffer on the pad just modifies the name to show this final
    graph point and the prints it to prove it all works.
    """
    super().pull(pad, bufs)
    # bufs = self.preparedframes[pad]
    # FIXME: use preparedframes
    self.cnt[pad] += 1
    impulse_offset = bufs.metadata["impulse_offset"]
    self.impulse_offset = impulse_offset
    # if bufs.EOS:
    #    self.mark_eos(pad)
    padname = pad.name.split(":")[-1]
    if padname == self.data_pad:
        if bufs.buffers is not None:
            if self.verbose:
                print(self.cnt[pad], bufs)
        for buf in bufs:
            if buf.end_offset > impulse_offset:
                if buf.offset < impulse_offset + Offset.fromsec(
                    self.template_duration + 1
                ):
                    # only save data around the impulse
                    buf.data = buf.data[self.bankno]
                    self.A.push(buf)
                else:
                    recovered_impulse_offset, match = self.impulse_test()
                    print(f"{impulse_offset=} {recovered_impulse_offset=} {match=}")
                    if impulse_offset == recovered_impulse_offset and match > 0.997:
                        print("Impulse test passed")
                    else:
                        print("Impulse test failed")
                    self.mark_eos(pad)
            else:
                # not at the impulse yet
                pass
    elif padname == self.impulse_pad:
        for buf in bufs:
            if buf.offset > impulse_offset - Offset.fromsec(
                1
            ) and buf.offset < impulse_offset + Offset.fromsec(
                self.template_duration
            ):
                # only save data around the impulse
                self.Ainput.push(buf)