Because we are allocating objects totaling over 80% of the ESP8266 memory
space, we are very sensitive to memory fragmentation. While we can control
our own allocs, the IP and SDK can allocate any size, at any time, which
can cause us to not be able to find the large contiguous memory areas we need
for the codes.
Work around this by preallocating, before anything else, buffer memory and
never freeing it. Pass this memory into the codec and it will allocate from
this block and be guaranteed to always succeed.
This may fix the last of the hangs on the webradio/http streaming, which only
seemed to happen after a stop/start cycle.
Check for NULL input data buffers and set values to NULL after freeing()
local variables to avoid memory corruption on use-after-freee.
Add lots of error injection in the host testing to help find bugs on
the PC where they're easier to debug.
When the input file(stream) doesn't open, you could end up calling
mp3::stop() which would try and mad_finish structures that were
not yet initialized. Now carry a flag when these are set, and only
_finish when it's set.
Valgrind found another uninitialized use. BitsPerSample was not being
called by the LIBMAD decoder, so it was possible to start out in a bad
state. Now always set it to 16, the decoder native output size.
Channels are initialized to 2, but this setting is always reset on
a per-frame basis anyway.
We removed the MP2-layer I and II decoders to save space, but an occasional
bad frame can come in that, by random chance, specifies one of this type.
Was calling NULL because of that. Now, just skip the frame and no crash.
The original I2S setup had the wrong # of bits being sent to the DAC
as well as wrong timeouts specified on the write_sample wrappers.
Fix those, allowing things like WebRadio.ino to work.
AAC-SBR allows for lower bitrates to have higher quality audio output,
and is used by many web radio stations. It was disables on the ESP8266
due to RAM limitations (about 50KB too little), but the ESP32 has no
such problem. Enable it automatically during compile.
The DATA chunk has a size field that was ignored. When there was additional
metadata after the chunk, however, it would try and play it leading to
clicks. Store and only read that amount of data now.
Emil Plalan found and fixed a bug where the buffer refill would loop
forever (until WDT) when there was no more data to read. For web streams
this wasn't detected, since there isn't really an EOF. However, single
files streamed from a web server would have this problem.
Because the mp3 player is a long living object and needs to allocate
large contiguous chunks on each frame, it is safer to allocate the
large xr[] and tmp[] arrays at the beginning. Add them to the frame
structure which lives the entire time the mp3 object is alive. No
heap allocations are now used by the mp3 player.
Move the imdct_s[] array to progmem, but do a memcpy_P() to stack
as the macros may cause a 16-bit access after optimization. It's only
36 words long and each data is read, anyway, so this doesn't increase
the CPU work significantly.
As seen in issue #21, the MP3 decoder would barf on a single lost sync
error, even though it was not actually a fatel error.
Rewrote the logic to include another Input() on DNF failure, and only die
when Input() actually fails. Seems to fix the problem in issue #21 and also
did not break the WebRadio.
DecodeNextFrame had a leftover while(1) in the inner loop without any
additional Input() calls, which could lead to an infinite loop on an
error condition. Remove that unneeded loop. Addresses some of the
stuff found in issue #21
Add handing of ID3V2.2.x (tested) and (I think) V2.3.x, and fix the
unsync handling.
Issue 23 found a problem where ID3V2.2 headers, which are very different
from v2.3 and v2.4, caused crashes. Fixing that showed another bug where
the unsyunc flag was not honored and all data was assumed to be unsync'd
causing improper reads of ID3 tags in some circumstances.
The ICYMDReader was not properly returning the unread "'" characters
and so StreamTitles were being truncated at the first apostrophe. Fix
the unread() and read() member functions to clean up.
@armSeb found that the stream interface caused a big slowdown and
hiccups in his playback, replace the MetadataCallback's *Stream with
a simple *char for faster operation and less overhead.
Include the RealNetworks HELIX MP3 decoder as AudioGeneratorMP3a.
Right now runs fine, but requires a full 1152*2 sample buffer and
seems to run slower than real time.
All objects can have a metadata and/or status report callback once created.
They will call the source program with the parsed ID3 tag, ICY StreamTitle,
or text and numeric error ID of any interesting event. It is up to the end
application to figure out what to do with this. Normally, you'd just print
it to the LCD or Serial port. But it is possible to detect, say, a buffer
overflow and then switch to a lower bitrate stream for example.
Allow a UI to get a notification when a StreamTitle or other metadata
comes into the stream. Callback function would normally just update a
flag and let the next global event loop handle updating the display.
Forgot to save the passed in *AudioFileSource, since it was not needed
in the MIDI playback engine (it's encapsulated in a buffered object). Now
with ::loop in every object, it's needed and we were calling a garbage
object causing a random crash.
ID3 tags can now automatically give your app a callback with different tag
types without any additional work on your application. See the PlayMP3 example
for a very simple example of AudioFileSourceID3's use.
Adds a virtual loop() method to the file input and audio output objects.
By default does nothing, but it can be used to prefill buffers/etc. in the
case of a buffered input, or to update a volume meter on an output, etc.
The AudioGenerators all now call it as part of their own loop()s, so there
is no user intervention required to make it work.
Thanks to @armSeb for the idea with his SPI RAM code.
The SBR code compiles, but it requires an additional 50,788 bytes of system
RAM to run so it's not going to work on the ESP8266. Add the code in just in
case someone wants to move it to another architecture w/more RAM available.
Wasn't incrementing the returned data pointer when we had to split the
requested read into pre- and post-ICY blocks. This resulted in errors
in the decoded stream.
Add a subclass of the HTTPStream that can extract the ICY metadata.
Needs to implement a callback function or parse it internally and expose
a class variable to let the UI see the current stream information.
@armSeb founf another case in issue #9 where the buffer would never
refill if the input stream was very slightly below the output stream
velocity. In that case, when we hit a 0-water mark, force a complete
refill of the buffer on the next pass.
MP3 maximum frame size of 1441 bytes, and now that there is a real
buffering class available it makes no sense to include additional space
in ths decoder. Reduce to the minimum to save memory in default case.
A simple FIR filter with decimation and specifable taps and rational
decimation allows for frequency conversion or any other kind of filtering
desired.
The buffer class seems to work very well to avoid hiccups while
streaming from the web, so add it by default to the example to
make it easier for users to cut-n-paste and make it work.
@armSeb, in issue 7, noticed noise being introduced when amplifying the
output. By moving the amplification to right before submitting the sample
to I2S it went away. Update with his suggested fix.
Correct issue @armSeb found and fixed where some bytes on the fallthrough
case were not being recorded as stored, leading to mp3 playback errors.
Fixes one half of issue 7.