Fix default_value for OpenAI voice dropdown

[petenelson]

While working with a fresh installation of ClassifAI, I added the API key for OpenAI Text to Speech and saved it, then tried generating text-to-speech for a post, but it was failing with their API returning the following error:

"[{'type': 'enum', 'loc': ('body', 'voice'), 'msg': "Input should be 'nova', 'shimmer', 'echo', 'onyx', 'fable', 'alloy', 'ash', 'sage' or 'coral'", 'input': 'voice', 'ctx': {'expected': "'nova', 'shimmer', 'echo', 'onyx', 'fable', 'alloy', 'ash', 'sage' or 'coral'"}}]"

After some debugging, I found that that the default_value for the voice selection was set to voice, and it reflected this in the classifai_feature_text_to_speech_generation option:

'openai_text_to_speech' =>
array (
  'api_key' => 'redacted',
  'authenticated' => true,
  'tts_model' => 'tts-1',
  'voice' => 'voice',
  'format' => 'mp3',
  'speed' => '1',
),

After changing the voice value to something else, saving it, then setting it to Alloy and saving it again, the the correct value showed up for voice:

'openai_text_to_speech' =>
array (
  'api_key' => 'redacted',
  'authenticated' => true,
  'tts_model' => 'tts-1',
  'voice' => 'alloy',
  'format' => 'mp3',
  'speed' => '1',
),

And generating text-to-speech for the post was working correctly. This PR sets a valid value for the default voice.

[github-actions]

@petenelson thanks for the PR! Could you please fill out the PR template with description, changelog, and credits information so that we can properly review and merge this?