Compiling pipper with pcaudiolib

[StoryHack]

I'm attempting to generate several .wavs using the json lines format, with "output_file" set in each json line. When I attempt to run piper using my .jsonl file, I get an error that says:

WARNING: Piper was not compiled with pcaudiolib. Output audio will be written to the current directory.

And all the wav files are indeed created with automatically generated names in the current directory. But I would really like to put the wavs with the names that get put in the directories I want. So, I'm tring to figure out how to compile piper with pcaudiolib.

I did put the piper-phonemize stuff in the correct lib/ location. And I have installed a bunch of compiler-related packages according to the errors that running "make" was giving me. Now, when I run make, get the following output

> make
mkdir -p build
cd build && cmake ../src/cpp -DCMAKE_BUILD_TYPE=Release && make
-- Configuring done
-- Generating done
-- Build files have been written to: /home/bryce/piper/build
make[1]: Entering directory '/home/bryce/piper/build'
make[2]: Entering directory '/home/bryce/piper/build'
make[3]: Entering directory '/home/bryce/piper/build'
Consolidate compiler generated dependencies of target piper
make[3]: Leaving directory '/home/bryce/piper/build'
make[3]: Entering directory '/home/bryce/piper/build'
[ 33%] Linking CXX executable piper
/usr/bin/ld: cannot find -lespeak-ng: No such file or directory
/usr/bin/ld: cannot find -lonnxruntime: No such file or directory
collect2: error: ld returned 1 exit status
make[3]: *** [CMakeFiles/piper.dir/build.make:113: piper] Error 1
make[3]: Leaving directory '/home/bryce/piper/build'
make[2]: *** [CMakeFiles/Makefile2:83: CMakeFiles/piper.dir/all] Error 2
make[2]: Leaving directory '/home/bryce/piper/build'
make[1]: *** [Makefile:91: all] Error 2
make[1]: Leaving directory '/home/bryce/piper/build'
make: *** [Makefile:9: piper] Error 2

I did install espeak and the onnxruntime, and I'm still gettig the same error. And I still don't know how to add in pcaudiolib.

What am I doing wrong?

[synesthesiam]

Did you run piper with the --json-input flag?

pcaudiolib is only used to play audio directly to the speakers. I don't compile it in by default because it pulls in a lot of other dependencies.

[StoryHack]

Here is the command:

cat ./tmp/cues.jsonl | ./piper/piper --model voice/libritts.onnx --json-input

And my jsonl file looks like:

{ "text": "Chapter 1", "speaker_id": 0, "output_file": "./tmp/1.wav" }
{ "text": "This is a test of json input.", "speaker_id": 1, "output_file": "./tmp/2.wav" }
{ "text": "I really hope I can get it work work.", "speaker_id": 1, "output_file": "./tmp/3.wav" }

The wavs are generated as if they are just lines of plain text, as I can hear it say "text" then the line, then some gibberishy stuff ending in "wav"

I get similar results when I try to echo a single line of json.

echo '{ "text": "Welcome again to the world of speech synthesis!"}' | ./piper/piper --model voice/libritts.onnx speaker-id 0 --output_file welcomeagain.wav --json-input

I must be doing something wrong, but for the life of me, I can't figure out what.

[StoryHack]

Nevermind. Once again, I'm an idiot. Somehow I hadn't updated to the latest binary release. Once I did, everything worked as I wanted, and now I have no need of compiling. Thanks for putting up with me.

[synesthesiam]

No worries, glad it's working!