start/stoping AnalogAudioStream and AudioRealFFT #755
-
|
Hello, void setupFFT(int sr) { void StartFFT(int sr) { void StopFFT() { void setupGeneralAdc(int pin) { void StartAdc() { void StopAdc() { |
Beta Was this translation helpful? Give feedback.
Replies: 5 comments 9 replies
-
|
You would need to figure out why it is crashing: analyse the stacktrace ! Are you sure that you do not write any data to the closed fft ? |
Beta Was this translation helpful? Give feedback.
-
|
Analyzing the crashdump exeeds my capabilities.
I use AudioRealFFT
It would already be a help when you could confirm how to properly start/stop AudioStreams and fft. I’m not sure if the ..end() method is the right way to do. E.g. if ..end() destroys all objects and allows me to create new ones during runtime
… On 6 Apr 2023, at 18:05, Phil Schatzmann ***@***.***> wrote:
You would need to figure out why it is crashing: analyse the stacktrace !
You did not mention which of the many FFT implementation you were selecting...
—
Reply to this email directly, view it on GitHub <#755 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AOWLL6QNNKIPVEHS2LCBE4LW73SU3ANCNFSM6AAAAAAWVTAXNQ>.
You are receiving this because you authored the thread.
|
Beta Was this translation helpful? Give feedback.
-
|
I committed a correction that might help. |
Beta Was this translation helpful? Give feedback.
-
Beta Was this translation helpful? Give feedback.
-
|
The Exception decoder can not decode my dump. I'm sure the mistake is on my side, I use this the first time. hange Mode to: 4 Core 1 register dump: Backtrace: 0x400d59e9:0x3ffcb290 0x400d5f1b:0x3ffcb2b0 0x400d5fca:0x3ffcb2d0 0x400d6047:0x3ffcb2f0 0x400d51a2:0x3ffcb310 0x400d545f:0x3ffcb350 0x400d48bb:0x3ffcb380 0x400d4994:0x3ffcb3a0 0x400d6a8d:0x3ffcb3d0 0x400da5d9:0x3ffcb3f0 ELF file SHA256: 24811b65e718b4d9 Rebooting... |
Beta Was this translation helpful? Give feedback.
-
|
Did you update the library? I can't help you with the decoding because the decoder needs the compilation result to do the analysis. |
Beta Was this translation helpful? Give feedback.
-
|
I would think so. I have 2.0.7 for esp32 and downloaded the latest versions for audio-tools, RealFFT and A2DP from your github. In the meantime I figured out how to decode the crash-dump. I did miss installing ..libcurses.. C: 0x400d59e4: ffft::FFTReal ::compute_direct_pass_1_2(float*, float const*) const at .../Projects/sketchbook.lx/libraries/arduino-audio-tools/src/AudioLibs/FFT/FFTReal.h line 1067 Decoding stack results |
Beta Was this translation helpful? Give feedback.
-
|
I was just running a test and for me this is working w/o any issues. Are you sure that you installed the latest version when I told you that I have committed a correction ? Here is my test case: #include "AudioTools.h"
#include "AudioLibs/AudioKit.h"
#include "AudioLibs/AudioRealFFT.h" // or AudioKissFFT
AudioKitStream kit; // Audio source
AudioRealFFT fft; // or AudioKissFFT
StreamCopy copier(fft, kit); // copy mic to tfl
int channels = 2;
int samples_per_second = 44100;
int bits_per_sample = 16;
float value=0;
// display fft result
void fftResult(AudioFFTBase &fft){
float diff;
auto result = fft.result();
if (result.magnitude>100){
Serial.print(result.frequency);
Serial.print(" ");
Serial.print(result.magnitude);
Serial.print(" => ");
Serial.print(result.frequencyAsNote(diff));
Serial.print( " diff: ");
Serial.println(diff);
}
}
void setup() {
Serial.begin(115200);
AudioLogger::instance().begin(Serial, AudioLogger::Info);
// setup Audiokit
auto cfg = kit.defaultConfig(RX_MODE);
cfg.input_device = AUDIO_HAL_ADC_INPUT_LINE2;
cfg.channels = channels;
cfg.sample_rate = samples_per_second;
cfg.bits_per_sample = bits_per_sample;
kit.begin(cfg);
// Setup FFT
auto tcfg = fft.defaultConfig();
tcfg.length = 8192;
tcfg.channels = channels;
tcfg.sample_rate = samples_per_second;
tcfg.bits_per_sample = bits_per_sample;
tcfg.callback = &fftResult;
fft.begin(tcfg);
copier.copyN(100);
fft.end();
fft.begin(tcfg);
}
void loop() {
copier.copy();
} |
Beta Was this translation helpful? Give feedback.
-
|
Mea Culpa, one one hand: I did not realize that you just recently updated the audio library. I did update it this morning and my sketch runs much more stable. E.g. I can switch between vaious fft configs (only the sampling rate changes). I can issue fft.begin(config1) .. fft.end().. fft.begin(config2).. fft.end() etc.Big step forward. Thanks for looking into this. BUT. First of all, my work-around with the older library was to just skip the fft.end() part. That seemed wrong but it made my sketch running perfectly fine. Now, after updating the library I still can provoke crashes. It happens after I stoped FFT with fft.end() and wanted to start a new one (setupFFT(scan rate) which is basically the same code as your FFT setup). The stack trace looks like listed below. Please note also that your example does not reflect my use-case. I have three input channels (two internal ADC, one bluetooth A2DP). Each has different sampling rate so when I switch between the input channels, I stop fft and start it again with the new config. Consequently, I have three copier objects that I switch between. Decoding stack results |
Beta Was this translation helpful? Give feedback.
-
|
You are just running out of RAM! You could resample to keep the sampling rate consistent. I also think that the different FFT implementations have different RAM requirements, so you could try to compare them in this regard. Or you could use a ESP32 which provides PSRAM. |
Beta Was this translation helpful? Give feedback.
-
|
Good point. I thought of that as well and I was not sure how that is handled. Obviously there is no memory management. But what happens during run-time when I start/stop FFT ? Is at start (fft.begin()) just memory allocated on the heap which will be freeed on fft.end() ?If this is the case, my sequence of fft.begin()...fft.end() should not consume condiderabely more RAM than just one call of fft.begin(). Anyway, having just one fft.begin() and one constante sample rate is defenitely doable for me and might even be the cleaner code. |
Beta Was this translation helpful? Give feedback.
-
|
Yes, start() allocates memory both for the input & result array on the heap and calls the fft specific initialisation and end() is releasing it again. |
Beta Was this translation helpful? Give feedback.
-
|
hmm, well. Speaking of resampling. Did I miss anything basic ? I can not compile even the simple resample examples you provide. I get at compile time when declaring something like ResampleStream<int16_t> resample(sound); |
Beta Was this translation helpful? Give feedback.
-
|
There were some changes and I might not have updated the examples: |
Beta Was this translation helpful? Give feedback.
I committed a correction that might help.
I did not reset the pointers to null so when the FFT object was created after calling end the necessary arrays and objects were not allocated any more...