Crash_Override/FreeDATA
1
0
Fork 0
mirror of https://github.com/DJ2LS/FreeDATA synced 2024-05-14 08:04:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

modem optimization

Browse source 
optimized chunk size for audio data. DATAC3 should now work and overall SNR is optimized since we have less biterrors beacuse of losses by simultaneous buffer allocation
This commit is contained in:
DJ2LS 2021-10-03 16:31:34 +02:00
parent 0b115bd00c
commit e7ccf42ec3
2 changed files with 50 additions and 38 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

82
tnc/modem.py
View file

@ -63,7 +63,14 @@ class MODEMSTATS(ctypes.Structure):
class RF():
def __init__(self):
self.AUDIO_SAMPLE_RATE_RX = 48000
self.AUDIO_SAMPLE_RATE_TX = 48000
self.MODEM_SAMPLE_RATE = 8000
self.AUDIO_FRAMES_PER_BUFFER = 8192
self.AUDIO_CHUNKS = 48 #8 * (self.AUDIO_SAMPLE_RATE_RX/self.MODEM_SAMPLE_RATE) #48
self.AUDIO_CHANNELS = 1
# -------------------------------------------- LOAD FREEDV
try:
# we check at first for libcodec2 in root - necessary if we want to run it inside a pyinstaller binary
@ -81,17 +88,17 @@ class RF():
atexit.register(self.p.terminate)
# --------------------------------------------OPEN AUDIO CHANNEL RX
self.stream_rx = self.p.open(format=pyaudio.paInt16,
channels=static.AUDIO_CHANNELS,
rate=static.AUDIO_SAMPLE_RATE_RX,
frames_per_buffer=static.AUDIO_FRAMES_PER_BUFFER,
channels=self.AUDIO_CHANNELS,
rate=self.AUDIO_SAMPLE_RATE_RX,
frames_per_buffer=self.AUDIO_FRAMES_PER_BUFFER,
input=True,
input_device_index=static.AUDIO_INPUT_DEVICE
)
# --------------------------------------------OPEN AUDIO CHANNEL TX
self.stream_tx = self.p.open(format=pyaudio.paInt16,
channels=1,
rate=static.AUDIO_SAMPLE_RATE_TX,
frames_per_buffer=static.AUDIO_FRAMES_PER_BUFFER, # n_nom_modem_samples
rate=self.AUDIO_SAMPLE_RATE_TX,
frames_per_buffer=self.AUDIO_FRAMES_PER_BUFFER, # n_nom_modem_samples
output=True,
output_device_index=static.AUDIO_OUTPUT_DEVICE, # static.AUDIO_OUTPUT_DEVICE
)
@ -262,7 +269,7 @@ class RF():
self.streambuffer += bytes(mod_out)
self.streambuffer += bytes(mod_out_postamble)
converted_audio = audioop.ratecv(self.streambuffer, 2, 1, static.MODEM_SAMPLE_RATE, static.AUDIO_SAMPLE_RATE_TX, None)
converted_audio = audioop.ratecv(self.streambuffer, 2, 1, self.MODEM_SAMPLE_RATE, self.AUDIO_SAMPLE_RATE_TX, None)
self.streambuffer = bytes(converted_audio[0])
# append frame again with as much as in count defined
for i in range(1, count):
@ -351,7 +358,7 @@ class RF():
self.c_lib.freedv_rawdatapostambletx(freedv, mod_out_postamble)
self.streambuffer += bytes(mod_out_postamble)
converted_audio = audioop.ratecv(self.streambuffer, 2, 1, static.MODEM_SAMPLE_RATE, static.AUDIO_SAMPLE_RATE_TX, None)
converted_audio = audioop.ratecv(self.streambuffer, 2, 1, self.MODEM_SAMPLE_RATE, self.AUDIO_SAMPLE_RATE_TX, None)
self.streambuffer = bytes(converted_audio[0])
# -------------- transmit audio
@ -439,7 +446,7 @@ class RF():
#pass
self.c_lib.freedv_set_frames_per_burst(freedv, 0)
'''
fft_buffer = bytes()
while True:
'''
@ -450,11 +457,11 @@ class RF():
'''
data_in = bytes()
data_in = self.stream_rx.read(1024, exception_on_overflow=False)
data_in = self.stream_rx.read(self.AUDIO_CHUNKS, exception_on_overflow=False)
#self.fft_data = data_in
data_in = audioop.ratecv(data_in, 2, 1, static.AUDIO_SAMPLE_RATE_RX, static.MODEM_SAMPLE_RATE, None)
data_in = audioop.ratecv(data_in, 2, 1, self.AUDIO_SAMPLE_RATE_RX, self.MODEM_SAMPLE_RATE, None)
data_in = data_in[0] # .rstrip(b'\x00')
self.fft_data = data_in
#self.fft_data = data_in
# we need to set nin * 2 beause of byte size in array handling
datac0_nin = self.c_lib.freedv_nin(datac0_freedv) * 2
@ -466,6 +473,11 @@ class RF():
datac0_buffer += data_in
datac1_buffer += data_in
datac3_buffer += data_in
# refill fft_data buffer so we can plot a fft
if len(self.fft_data) < 1024:
self.fft_data += data_in
# DECODING DATAC0
if len(datac0_buffer) >= (datac0_nin):
@ -682,26 +694,30 @@ class RF():
while True:
time.sleep(0.01)
# WE NEED TO OPTIMIZE THIS!
data_in = self.fft_data
if len(self.fft_data) >= 1024:
data_in = self.fft_data
self.fft_data = bytes()
# https://gist.github.com/ZWMiller/53232427efc5088007cab6feee7c6e4c
audio_data = np.fromstring(data_in, np.int16)
# Fast Fourier Transform, 10*log10(abs) is to scale it to dB
# and make sure it's not imaginary
# https://gist.github.com/ZWMiller/53232427efc5088007cab6feee7c6e4c
audio_data = np.fromstring(data_in, np.int16)
# Fast Fourier Transform, 10*log10(abs) is to scale it to dB
# and make sure it's not imaginary
try:
fftarray = np.fft.rfft(audio_data)
# set value 0 to 1 to avoid division by zero
fftarray[fftarray == 0] = 1
dfft = 10.*np.log10(abs(fftarray))
dfftlist = dfft.tolist()
# send fft only if receiving
if static.CHANNEL_STATE == 'RECEIVING_SIGNALLING' or static.CHANNEL_STATE == 'RECEIVING_DATA':
#static.FFT = dfftlist[20:100]
static.FFT = dfftlist
except:
print("setting fft = 0")
# else 0
static.FFT = [0] * 400
try:
fftarray = np.fft.rfft(audio_data)
# set value 0 to 1 to avoid division by zero
fftarray[fftarray == 0] = 1
dfft = 10.*np.log10(abs(fftarray))
dfftlist = dfft.tolist()
# send fft only if receiving
if static.CHANNEL_STATE == 'RECEIVING_SIGNALLING' or static.CHANNEL_STATE == 'RECEIVING_DATA':
#static.FFT = dfftlist[20:100]
static.FFT = dfftlist
except:
print("setting fft = 0")
# else 0
static.FFT = [0] * 400
else:
pass

6
tnc/static.py
View file

@ -56,11 +56,7 @@ SCATTER = []
AUDIO_INPUT_DEVICE = 1
AUDIO_OUTPUT_DEVICE = 1
AUDIO_SAMPLE_RATE_RX = 48000
AUDIO_SAMPLE_RATE_TX = 48000
MODEM_SAMPLE_RATE = 8000 # 8000
AUDIO_FRAMES_PER_BUFFER = 16384 # 256 # 512 # 1024 #2048 --> nicht 880 # 128 gut, 256, 1024 16384
AUDIO_CHANNELS = 1
AUDIO_RMS = 0
FFT = []