docker compose up --build Error : espeak-ng-data/phontab': No such file or directory.

[small-tk]

kokoro-tts-1 | Error processing file '/home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic/share/espeak-ng-data/phontab': No such file or directory.

[651764643]

This issue occurs on a computer with an x86 architecture, using an Intel CPU.

[TheMrCodes]

Still a problem for me on windows with AMD Ryzen processor.

@remsky But got it patch by changing the folder creation hierarchy, by first creating the share.

Here is my one line change:

diff --git a/docker/gpu/Dockerfile b/docker/gpu/Dockerfile
index 2360b82..86d041c 100644
--- a/docker/gpu/Dockerfile
+++ b/docker/gpu/Dockerfile
@@ -3,7 +3,8 @@ FROM --platform=$BUILDPLATFORM nvidia/cuda:12.4.1-cudnn-runtime-ubuntu22.04
 ENV DEBIAN_FRONTEND=noninteractive
 
 # Install Python and other dependencies
-RUN apt-get update && apt-get install -y \
+RUN mkdir -p /usr/share/espeak-ng-data \
+  && apt-get update && apt-get install -y \
     python3.10 \
     python3.10-venv \
     espeak-ng \

[remsky]

Feel free to submit a PR, but otherwise thanks! Will take a look later tonight. I see I did partially miss the symlink to handle all cases which could have been part of the cause.

[shinishiho]

I'm having this issue running with uv on macOS as well

[dong4j]

@shinishiho I encountered the error while running on M2, but @TheMrCodes method is also applicable to macOS. First, execute mkdir -p /usr/share/espeak-ng-data.

[thomas-hensel]

On Mac M3 via OrbStack - I started with the docker cmd ...
and got the error
Error processing file '/home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic/share/espeak-ng-data/phontab': No such file or directory.

I created in OrbStack Files section
/home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic
set permissions world readable
started a shell and soft linked /usr/share into /home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic

This has worked for me for short texts <200 characters

Longer texts bring: Error generating speech: The object is in an invalid state.

[remsky]

Taking a look, it looks like the issue isn't consistent between systems, a conflict with the upstream espeak-loader and the system version I install. Digging into to find a solution across all builds

[remsky]

hexgrad/misaki@000dd1e

In progress on the fix, think it's related to this commit from a few days ago.

[markldn]

This fixed it for me

mkdir -p /home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic/share/

ln -s /app/.venv/lib/python3.10/site-packages/espeakng_loader/espeak-ng-data /home/runner/work/espeakng-loader/espeakng-loader/espeak-ng/_dynamic/share/espeak-ng-data

[remsky]

https://github.com/remsky/Kokoro-FastAPI/releases/tag/v0.2.1

Should be implemented here, as far as I was able to test on a few machines. Let me know if the issues persists