README: added Docker build and run instructions

[rudolfolah]

• build the image with a tag name
• run the image, downloading voice files into container and copying output to host

[sebastiancarlos]

Thanks! This worked for me on a non-M1 mac.

The only issue for me is that the --output_file welcome.wav part doesn't seem to work. If I remove it, a file with a default name is generated in the current directory.

[rudolfolah]

Thanks! This worked for me on a non-M1 mac.

Awesome 😎

The only issue for me is that the --output_file welcome.wav part doesn't seem to work. If I remove it a file with a default name is generated in the current directory.

I wonder if using an absolute path would work or using --output_dir

[ishan0102]

Worked for me on an M1 mac! And similarly, --output_file didn't save it and --output_dir claimed to have saved it but I wasn't seeing it. No args worked fine, it saved to the working directory. Screenshots attached.

--output_file: doesn't say it is saved anywhere

--output_dir: says it is saved to welcome.wav dir but doesn't show up when running ls

no args: shows up as 1688860855967686335.wav

[ishan0102]

Also quick question, does this mean piper can only be called by spinning up a docker image and running these commands? Is there a way to get piper into my path so that I can call it from anywhere? I'm trying to get this repo working on my M1 mac: https://github.com/yacineMTB/talk

[rudolfolah]

based on the comments suggesting that the --output_file and --output_dir command line arguments are not working as expected (could be working and putting it in the wrong path?), here's a possible workaround for the current directory:

Suggested change

	echo 'Welcome to the world of speech synthesis!' | \
	./piper --model en-us-lessac-medium.onnx --output_file welcome.wav
	# while the container is running, and in a different terminal, copy the output file to the host
	docker cp $(docker ps -lq):/dist/piper/welcome.wav .
	echo 'Welcome to the world of speech synthesis!' | \
	./piper --model en-us-lessac-medium.onnx
	# while the container is running, and in a different terminal, copy the output files to the host
	docker exec $(docker ps -lq)bash -c "ls /dist/piper/*.wav" | while read line; do docker cp $(docker ps -lq):/$line .; done

based on this StackOverflow answer, I have not tested this yet

[ishan0102]

Kept running into errors with this along the lines of:

However I was able to get a quick bash script that accomplishes copying all of the files over:

#!/bin/bash

# Get the ID of the most recently created Docker container
container_id=$(docker ps -lq)

# Execute a command in the Docker container to get the list of .wav files
files=$(docker exec -it $container_id bash -c "cd /dist/piper && ls *.wav")

# Remove any carriage return characters (in case this script is run on Windows)
files=${files//$'\r'/}

# Loop over each file
for file in $files; do
  # Copy the file from the Docker container to the current directory
  docker cp "$container_id:/dist/piper/$file" .
done

*script generated by GPT-4, works on my machine but someone should probably verify this

[rudolfolah]

that script works; on Windows I had to run bash first

[ishan0102]

sweet

[rudolfolah]

here's an alternative way of running piper in Docker where the output directory is mounted:

mkdir output

# Linux, MacOS
docker run -it --name piper --entrypoint /bin/bash -w /dist/piper --volume $PWD:/dist/piper/output -d piper:latest
# Windows
docker run -it --name` piper --entrypoint /bin/bash -w /dist/piper --volume ${PWD}/output:/dist/piper/output -d piper:latest

docker exec piper bash -c 'curl -L --output voice-en-us-lessac-medium.tar.gz https://github.com/rhasspy/piper/releases/download/v0.0.2/voice-en-us-lessac-medium.tar.gz && tar -xvf voice-en-us-lessac-medium.tar.gz'

docker exec piper bash -c "echo 'Welcome to the world of speech synthesis!' | ./piper --model en-us-lessac-medium.onnx --output_dir ./output"

Output dir and the mounted volume worked for me on Windows:

[sebastiancarlos]

here's an alternative way of running piper in Docker where the output directory is mounted:

Works great for me, thanks!. I just had to add the missing /output in the Linux/MacOS version after $PWD. And I removed the exclamation mark from the sample text, which hilariously crashes on Bash.

[KarlLivesey]

I'm cloning the repo and am getting this when running the docker build what am I missing

docker buildx build --target build -t piper:latest .
[+] Building 0.3s (12/15) docker:desktop-linux
=> [internal] load .dockerignore 0.0s
=> => transferring context: 170B 0.0s
=> [internal] load build definition from Dockerfile 0.0s
=> => transferring dockerfile: 2.53kB 0.0s
=> [internal] load metadata for quay.io/pypa/manylinux_2_28_aarch64:latest 0.1s
=> [build-arm64 1/1] FROM quay.io/pypa/manylinux_2_28_aarch64@sha256:9484084f91da3883b7c79aeab5cee5a466abd5a34fe9cfb243a26d5e8534a6bb 0.0s
=> [internal] load build context 0.0s
=> => transferring context: 996.82kB 0.0s
=> CACHED [build 1/10] WORKDIR /build 0.0s
=> CACHED [build 2/10] RUN curl -L "https://github.com/gabime/spdlog/archive/refs/tags/v1.11.0.tar.gz" | tar -xzvf - && mkdir -p "spdlog-1.11.0/build" && cd "spdlog-1.11.0/build" && cmake .. && make -j8 && cmake --install . --prefix /usr 0.0s
=> CACHED [build 3/10] RUN mkdir -p "lib/Linux-$(uname -m)" 0.0s
=> CACHED [build 4/10] RUN mkdir -p "lib/Linux-$(uname -m)/piper_phonemize" && curl -L "https://github.com/rhasspy/piper-phonemize/releases/download/v1.1.0/libpiper_phonemize-arm64.tar.gz" | tar -C "lib/Linux-$(uname -m)/piper_phonemize" -xzvf - 0.0s
=> CACHED [build 5/10] COPY VERSION Makefile ./ 0.0s
=> CACHED [build 6/10] COPY src/cpp/ ./src/cpp/ 0.0s
=> ERROR [build 7/10] RUN make 0.1s

[build 7/10] RUN make:
0.087 cmake -Bbuild -DCMAKE_INSTALL_PREFIX=install
0.101 CMake Error: The source directory "/build" does not appear to contain CMakeLists.txt.
0.101 Specify --help for usage, or press the help button on the CMake GUI.
0.104 make: *** [Makefile:4: all] Error 1

---

Dockerfile:44

42 | COPY VERSION Makefile ./
43 | COPY src/cpp/ ./src/cpp/
44 | >>> RUN make
45 |
46 | # Do a test run

ERROR: failed to solve: process "/bin/sh -c make" did not complete successfully: exit code: 2

running on m2 max