Capture audio doubt - How buffer is filled #1938
-
|
Hello @pschatzmann, I have a doubt about read function to recover capture samples from ES8388. For example: The READ function wait for fill buffer complete (until samples defined by BUFFER_SIZE_REC) or when I call READ, get until then all bytes that it could capture? Is there any example that uses that function and later processed all content buffer, on continuos capturing, without samples lossing? Thanks in advance. |
Beta Was this translation helpful? Give feedback.
Replies: 6 comments 39 replies
-
|
Please read the first Introduction Wiki: there is no read(bufferRecCopy, BUFFER_SIZE_REC); method! The Method from the Arduino Stream API is readBytes(buffer, size) and provides the requested bytes (or to be avoided read() which provides one single byte). If is used in all examples as part of the StreamCopy as descirbed in the Wiki |
Beta Was this translation helpful? Give feedback.
-
|
Please explain what you are trying to do: I don't understand "The 8-bit machine sends audio pulses". The aux pin of the audio board can read an analog line level audio signal and the microphone can capture sound. You seem to do something else... |
Beta Was this translation helpful? Give feedback.
-
|
Ok. I'm trying read data from ZX Spectrum machine through audio capture with AUDIOKIT (E8388) When the audio is in the buffer I'm going to read the value of each sample and counting these to determine if the original pulses are high or low, and how the wide is each one. There are four differents kind of pulses. If the sampling rate is at 44.1KHz then
But I've observed that captures audio is not same through Audio-kit (sending the buffer data to output directly to line-out) versus directly from 8-bit computer, using the same sound card in both case to compare signals. But I'm creating audio pulses with similar way to play TAP or TZX files through audio-kit to ZX Spectrum and all works fine (sampling rate at 44.1KHz, 16 bits mono, left channel) Then I don't understand why pulses are differents when the audio-kit capture these. |
Beta Was this translation helpful? Give feedback.
-
|
I still don't quite understand: why are you reading a binary singal via the ADC ? |
Beta Was this translation helpful? Give feedback.
-
|
Ok Phil. My project, PowaDCR is based on AUDIOKIT A1S board by AI Thinker. The idea is to use the board without additional hardware, then. AUDIOKIT board has, one line-in, one line-out and one two single channels amplified out. Then if I use the digital pins users have to prepare a jack to connect at this pins and this solution is out of my project. Several applications to convert WAV files to TAP or TZX files work with ADC of the sound card on PC, then this project can convert on real time audio been captured from the ZX Spectrum without problems. My main problem now is maybe I don't using correctly your library, then I'd like to receive support by you. Thanks in advance |
Beta Was this translation helpful? Give feedback.
-
|
Please share how you are currently reading the samples and how you were setting up I2S. I don't know if this will work conceptionally, since I never tried to read a binary signal via the ADC. |
Beta Was this translation helpful? Give feedback.
-
|
I'm going to try to do some more tests. Can these includes coexist together? When I try to use "AudioBoardStream" compiler out a lot of errors, even removing the Audiokit.h, too. Do I need anything else additional to the AudioBoardStream installed on my project (I use VSCode and platformIO) to detect drivers or something about? |
Beta Was this translation helpful? Give feedback.
-
|
Well, Doing test with audio-kit capturing audio using a PC, Sound Blaster G6 and Audacity: I would to know why the audio captured is different through the audio-kit (from line-in to line-out sample by sample without postprocessing) vs captured directly to the audio source without audiokit. Finally, pulses that were captured are differents for each testing scenery. Note: the audio in both testing scenery is noise free. About conversion yes, my project currently playing TAP and TZX, but not only for ZX Spectrum, also playing for MSX or Amstrad and other 8-bit machine with its specific formats of cassettes preservations files. Yes I could to send you a WAV with pulses. My device generates WAV file of the audio output, too in order to know without external capturing device what pulses were generated. Thanks |
Beta Was this translation helpful? Give feedback.
-
|
Since you did not provide me with a wav file with tap impulses, I just cobbled together the following simple sketch and connected the Serial tx line with one of the audio inputs : #include "AudioTools.h"
#include "AudioTools/AudioLibs/AudioBoardStream.h"
AudioInfo info(44100, 2, 16);
I2SCodecStream in(AudioKitEs8388V1);
int16_t samples[2] = { 0, 0 };
void setup() {
Serial.begin(115200);
Serial1.begin(115200);
AudioToolsLogger.begin(Serial, AudioToolsLogLevel::Warning);
// setup input
auto cfg = in.defaultConfig(RX_MODE);
cfg.copyFrom(info);
cfg.input_device = ADC_INPUT_LINE2;
in.begin(cfg);
// set max input volume
in.setInputVolume(0.3);
}
void loop() {
// read a frame (left and right sample)
int read = in.readBytes((uint8_t*)&samples, sizeof(samples));
assert(sizeof(samples)==read);
Serial.printf("%d, %d\n", samples[0], samples[1]);
// generate test signal
Serial1.println("test");
}This was giving the following output in the Serial Plotter
Here are my conclusions:
|
Beta Was this translation helpful? Give feedback.
-
|
@pschatzmann I already had a recorder version but was not fine and I decided to reprogram this, but I had captured audio without problem through line-in. Then two possibilities:
I always used old version of AUDIOKIT library. Never had problems. In my last version I updated audiokit library, then I ask you try to found the possibly problem. Nowadays the project is working better with your obsolete version or no longer version, but capture is not faithful to the source, and I don't know the reason, and as I tell above, this occur since I updated AUDIOKIT library. |
Beta Was this translation helpful? Give feedback.
-
|
You are funny: the AudioKit library is not supported any more because Espressif does not support the old I2S API since Arduino ESP 3.0 core. The actual version is 3.1.3 with version 3.2 coming out soon... If you are still using an old 2.x release, you are several years behind! |
Beta Was this translation helpful? Give feedback.
-
|
Well @pschatzmann, These below are all libraries that my project uses. Please, tell me if the repository location is correct or not, and tell me where I can get the good repository, if not. Dependency Graph This is the contenct of platformio.ini platform = espressif32 @ 6.6.0 |
Beta Was this translation helpful? Give feedback.
-
|
I am sorry, but I don't follow the PlatformIO releases and you will need to find out yourself what IDF release is behind them. Arduino ESP32 3.2 -> IDF 5.4 Please note that it is the IDF that provides the basic I2S API and functionality. |
Beta Was this translation helpful? Give feedback.
-
|
I think that is the version installed but I'll forze to install again for ESP32dev board in platformIO. I'll tell you progression this night (Spain) Thank you so much. |
Beta Was this translation helpful? Give feedback.
-
|
Hello @pschatzmann . The message is:
|
Beta Was this translation helpful? Give feedback.
-
|
The same problem have with "Audio-tools @ 1.0.2" |
Beta Was this translation helpful? Give feedback.
-
|
The only way to update is not forzed to the version but versions downloaded are:
Not possible to download lastest versions. Why? |
Beta Was this translation helpful? Give feedback.
-
|
I'm already crazy. I have removed .pio directory and all dependecy setting from platformio.ini and manually download all your repositories with "git clone .." (as readme says to install for each one) and this is the result. Never the latest version are downloaded. Dependency Graph |
Beta Was this translation helpful? Give feedback.
-
|
Hello @pschatzmann I'm updating my code and using I2SCodecStream like a example that you recomend me. With 1.0.1 library compiler returns me the following message, #include "AudioTools.h" Can you help me? |
Beta Was this translation helpful? Give feedback.
-
|
I am not sure about the implications of allowing a copy of the I2SCodecStream: so I recommend to avoid to do this! |
Beta Was this translation helpful? Give feedback.
-
|
Hi @pschatzmann. At the end the issue was discovered. The source wave was generating as PWM signal. The only monitor that I had was Audacity. It seems that Audacity or my Sound capture board was converting PWM wave and I only was seeing a perfect square wave, but Audiokit was back crazy because was interpreting a PWM wave as normal square wave. Was the reason that wave was distorsioned. Now all is ok, and works fine, after I know this peculiarity of my equipment. I have been capturing some cassettes through old tape player and works fine. The source volume need a special attention due to be amplified but I'm happy with the results. Thanks for your support. |
Beta Was this translation helpful? Give feedback.
Please read the first Introduction Wiki: there is no read(bufferRecCopy, BUFFER_SIZE_REC); method!
The Method from the Arduino Stream API is readBytes(buffer, size) and provides the requested bytes (or to be avoided read() which provides one single byte).
If is used in all examples as part of the StreamCopy as descirbed in the Wiki