fix: fixed tts_server.py on macOS

[ErikBjare]

Important

Centralized audio output device selection in tts.py and improved espeak library detection for macOS in tts_server.py.

• Audio Output Device Selection:
  • Added get_output_device() in tts.py to centralize logic for selecting audio output devices.
  • Replaced inline device selection logic in audio_player_thread() and speak() with get_output_device().
• macOS Compatibility:
  • In tts_server.py, set espeak library path explicitly for macOS using EspeakWrapper.set_library().
  • Replaced subprocess with shutil.which() for espeak detection in _check_espeak().
• Misc:
  • Added comments for exception handling in tts.py when TTS extras are not installed.

^{This description was created by for ac3e2a0. It will automatically update as commits are pushed.}

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to 4eb7d88 in 1 minute and 13 seconds

More details

• Looked at 151 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 1 drafted comments based on config settings.

1. gptme/tools/tts.py:241

• Draft comment:
  Using hostapi == 2 to identify CoreAudio is unreliable as hostapi indices can vary. Consider using sd.query_hostapis() to find the correct index for CoreAudio.
• Reason this comment was not posted:
  Decided after close inspection that this draft comment was likely wrong and/or not actionable:
  The code has multiple fallbacks - first trying system default, then CoreAudio devices, then any output device. Even if the hostapi index is wrong, the code will still work through the third fallback. The hostapi check is just an optimization to prefer CoreAudio devices when available. The risk of the index being wrong seems low compared to the complexity of adding hostapi querying logic.
  The comment raises a valid technical point - hardcoding hostapi indices is indeed not ideal. The code could be more robust by properly detecting CoreAudio.
  While technically correct, the current approach is pragmatic given the multiple fallbacks. The benefit of adding hostapi detection logic doesn't justify the added complexity.
  Delete the comment. While technically valid, the current code handles device selection robustly through fallbacks, making this optimization suggestion not important enough to warrant a change.

Workflow ID: wflow_8r08XAHdgRMvpQgt

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 3.33333% with 29 lines in your changes missing coverage. Please review.

Project coverage is 69.66%. Comparing base (db212bd) to head (ac3e2a0).

✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
gptme/tools/tts.py	3.33%	29 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #418      +/-   ##
==========================================
- Coverage   70.09%   69.66%   -0.43%     
==========================================
  Files          70       70              
  Lines        5849     5862      +13     
==========================================
- Hits         4100     4084      -16     
- Misses       1749     1778      +29     

Flag	Coverage Δ
anthropic/claude-3-haiku-20240307	68.42% <3.33%> (-0.19%)	⬇️
openai/gpt-4o-mini	67.57% <3.33%> (-0.58%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ellipsis-dev]

👍 Looks good to me! Incremental review on e186316 in 20 seconds

More details

• Looked at 199 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 4 drafted comments based on config settings.

1. scripts/tts_server.py:84

• Draft comment:
  The use of shutil.which to check for espeak and espeak-ng is a good improvement for reliability. This ensures the executables are checked in the system's PATH.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The use of shutil.which is a more reliable way to check for the presence of executables, as it checks the system's PATH for the executable. This change improves the robustness of the _check_espeak function.

2. gptme/tools/tts.py:207

• Draft comment:
  The get_output_device function centralizes the logic for selecting an audio output device, reducing code duplication and improving maintainability. This is a good practice.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The get_output_device function is a good addition to centralize the logic for selecting an audio output device. This reduces code duplication and improves maintainability.

3. gptme/tools/tts.py:299

• Draft comment:
  Ensure that the exception handling in audio_player_thread and speak allows the program to continue running even if no output device is found. This prevents the entire application from crashing due to a missing audio device.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The get_output_device function is used in multiple places, which is a good practice to avoid code duplication. However, the exception handling in audio_player_thread and speak should ensure that the program can continue running even if no device is found.

4. gptme/tools/tts.py:362

• Draft comment:
  Using os.getenv to retrieve the GPTME_TTS_VOICE environment variable is a good practice for configuration flexibility, allowing users to set preferences without code changes.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The use of os.getenv to get the GPTME_TTS_VOICE environment variable is a good practice for configuration flexibility. This allows users to set their preferred voice without modifying the code.

Workflow ID: wflow_7XPfb8vo4Yr3DoWH

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[ellipsis-dev]

👍 Looks good to me! Incremental review on ac3e2a0 in 15 seconds

More details

• Looked at 190 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 3 drafted comments based on config settings.

1. scripts/tts_server.py:84

• Draft comment:
  The use of shutil.which to check for espeak or espeak-ng is a more robust and efficient method than using subprocess.run. Good improvement!
• Reason this comment was not posted:
  Confidence changes required: 0%
  The use of shutil.which is a more robust way to check for the presence of executables, replacing the previous subprocess call. This is a good change.

2. gptme/tools/tts.py:207

• Draft comment:
  The get_output_device function is a good refactoring to centralize the logic for selecting the audio output device. This improves code reuse and maintainability.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The PR introduces a new function get_output_device which is used in multiple places to determine the audio output device. This refactoring improves code reuse and maintainability.

3. scripts/tts_server.py:48

• Draft comment:
  The macOS-specific workaround for setting the espeak library path ensures compatibility on macOS systems. This is a necessary change for macOS support.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The PR adds a specific workaround for macOS to set the espeak library path. This is necessary for compatibility and should be noted.

Workflow ID: wflow_dgdwctbJgeKyPq8T

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[ErikBjare]

Might need to search for this path more intelligently in the future as new espeak versions are released