Convert AudioData stream to arrays of floats, and convert arrays of floats to AudioData by lucasw · Pull Request #84 · ros-drivers/audio_common

x-raw allows my mic to work- needs to be parameter

lucasw

committed

Not compiling yet, want to be able to decode audio and then process arrays of samples.

lucasw

committed

py example doesn't work

lucasw

committed

This compiles but results in glib/gstreamer errors when it runs.

lucasw

committed

It looks like only one buffer is received, but I'm not handling it correctly so the callback gets triggered repeatedly with the same buffer. It seems like I need something else in the pipeline to r…

lucasw

committed

This may be the raw data- but what about 8-bit vs. 16-bit samples? Do I have to know that and cast the data appropriately, or can I force conversion to expected format?

lucasw

committed

I thought I was getting good decoded data from the microphone, but now I only see five good samples and everything else is zero.

lucasw

committed

Now seeing reasonable waveform, and it is 16-bit, which the viewer has to handle

lucasw

committed

Passing data as -1.0 to 1.0 float now, so receiving node doesn't have to understand it (should be 0 to 1.0?)

lucasw

committed

More visually pleasing but perhaps not super useful fade out of old waveforms

lucasw

committed

Generating a float array signal, now need to get it into gst encoded AudioData.

lucasw

committed

Trying to convert ros messages of floating point arrays to gstreamer format, currently getting

lucasw

committed

Should be able to trace problem with debug on:

lucasw

committed

Trying out paused, but the next problem seems to be with pads? gstpad.c gst_pad_push_data error pushing events, return not-negotiated (Seen on debug level 6)

lucasw

committed

Trying out S16LE as the input type, but now back to 0:00:00.276624022 16823 0x224a000 WARN basetransform gstbasetransform.c:1414:gst_base_transform_setcaps:<filter> transform could not transform au…

lucasw

committed

Trying out audio resample, no success yet

lucasw

committed

Finall got it! The whole problem was not specifying the layout as a string. The input and output rates do look different though

lucasw

committed

Making messages debug level, also adjusting visualization. Next step is to approximate nes sound chip with A-D voices, have four ros nodes in control of each, and then they can be played individually.

lucasw

committed

Trying out nes style noise, but the waveform doesn't sound right

lucasw

committed

Wave encoding doesn't work, trying to fix that- if I put too many caps in it results in internal data error. Also trying to make noise sound more like online examples- I think when sample rate exce…

lucasw

committed

flac seems to work, though I got a crash once (maybe a different node though), and can't get rqt to load the config file

lucasw

committed

disabling wave. Flac works but the waveform is really different, it seems like there ought to be an auditory difference (or maybe I need the conversion to float to understand the decoded format bet…

lucasw

committed

Rename projects for consistency, but maybe float_to_audio should be audio_from_float to have the same prefix as the rest of the projects? Also deleted a lot of commented out code. Inability to kill…

lucasw

committed

The launch file doesn't work when I launch it- something is wrong with float_to_audio, but when I run it manually it does work:

lucasw

committed

Forgot to add this headers

lucasw

committed

Add options to disable parts of the float to audio launch file

lucasw

committed

View a spectrogram of the audio

lucasw

committed

Bigger nperseg shows more detail in the spectrogram

lucasw

committed

flake8 compliance

lucasw

committed

flake8 compliance for view.py

lucasw

committed

flake8 compliance in float_to_audio scripts

lucasw

committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Convert AudioData stream to arrays of floats, and convert arrays of floats to AudioData#84

Commits on May 19, 2023

x-raw allows my mic to work- needs to be parameter

Not compiling yet, want to be able to decode audio and then process arrays of samples.

py example doesn't work

This compiles but results in glib/gstreamer errors when it runs.

It looks like only one buffer is received, but I'm not handling it correctly so the callback gets triggered repeatedly with the same buffer. It seems like I need something else in the pipeline to r…

This may be the raw data- but what about 8-bit vs. 16-bit samples? Do I have to know that and cast the data appropriately, or can I force conversion to expected format?

I thought I was getting good decoded data from the microphone, but now I only see five good samples and everything else is zero.

Now seeing reasonable waveform, and it is 16-bit, which the viewer has to handle

Passing data as -1.0 to 1.0 float now, so receiving node doesn't have to understand it (should be 0 to 1.0?)

More visually pleasing but perhaps not super useful fade out of old waveforms

Generating a float array signal, now need to get it into gst encoded AudioData.

Trying to convert ros messages of floating point arrays to gstreamer format, currently getting

Should be able to trace problem with debug on:

Trying out paused, but the next problem seems to be with pads? gstpad.c gst_pad_push_data error pushing events, return not-negotiated (Seen on debug level 6)

Trying out S16LE as the input type, but now back to 0:00:00.276624022 16823 0x224a000 WARN basetransform gstbasetransform.c:1414:gst_base_transform_setcaps:<filter> transform could not transform au…

Trying out audio resample, no success yet

Finall got it! The whole problem was not specifying the layout as a string. The input and output rates do look different though

Making messages debug level, also adjusting visualization. Next step is to approximate nes sound chip with A-D voices, have four ros nodes in control of each, and then they can be played individually.

Trying out nes style noise, but the waveform doesn't sound right

Wave encoding doesn't work, trying to fix that- if I put too many caps in it results in internal data error. Also trying to make noise sound more like online examples- I think when sample rate exce…

flac seems to work, though I got a crash once (maybe a different node though), and can't get rqt to load the config file

disabling wave. Flac works but the waveform is really different, it seems like there ought to be an auditory difference (or maybe I need the conversion to float to understand the decoded format bet…

Rename projects for consistency, but maybe float_to_audio should be audio_from_float to have the same prefix as the rest of the projects? Also deleted a lot of commented out code. Inability to kill…

The launch file doesn't work when I launch it- something is wrong with float_to_audio, but when I run it manually it does work:

Forgot to add this headers

Add options to disable parts of the float to audio launch file

View a spectrogram of the audio

Bigger nperseg shows more detail in the spectrogram

flake8 compliance

flake8 compliance for view.py

flake8 compliance in float_to_audio scripts