enh: the assistant now produces wav audio input -> change the input param in the chat completion request with multimodal model

Signed-off-by: Julien Veyssier <[email protected]>

Author: julien-nc

lib/Service/OpenAiAPIService.php

@@ -560,7 +560,7 @@ public function createChatCompletion(
 						'type' => 'input_audio',
 						'input_audio' => [
 							'data' => $userAudioPromptBase64,
-							'format' => 'mp3',
+							'format' => 'wav',
 						],
 					],
 				],