From ffd8c6f2a5b6b8cf32c5a7121c20aa02375170c6 Mon Sep 17 00:00:00 2001
From: Hamza-Ayed <hamzaayedflutter@gmail.com>
Date: Fri, 22 May 2026 15:27:09 +0300
Subject: [PATCH] Update test_audio to ignore system instructions for TTS

---
 backend/public/test_audio.php | 49 ++++++++++++++++++++++++++---------
 1 file changed, 37 insertions(+), 12 deletions(-)

diff --git a/backend/public/test_audio.php b/backend/public/test_audio.php
index aa09477..9bc05a2 100644
--- a/backend/public/test_audio.php
+++ b/backend/public/test_audio.php
@@ -39,8 +39,8 @@ $systemPrompt = "You are a friendly customer service assistant. Speak in a warm,
 $testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";
 
 $candidateModels = [
-    'gemini-2.5-flash-preview-tts',
     'gemini-3.1-flash-tts-preview',
+    'gemini-2.5-flash-preview-tts',
     'gemini-2.0-flash',
     'gemini-2.5-flash',
     'gemini-3.5-flash',
@@ -51,17 +51,32 @@ $successfulModel = null;
 $audioResponse = null;
 
 foreach ($candidateModels as $model) {
-    echo "Testing model: models/{$model} ... ";
+    echo "Testing model: models/{$model} ...\n";
     
     $url = 'https://generativelanguage.googleapis.com/v1beta/models/' . $model . ':generateContent?key=' . $apiKey;
-    $payload = json_encode([
-        'contents' => [[
+    
+    // For TTS models, we do not send systemInstruction because they don't support it (Developer instruction not enabled)
+    // Instead we prepend it to the user message parts
+    $useSystemInstructionField = !str_contains($model, 'tts');
+    
+    $contents = [];
+    if (!$useSystemInstructionField) {
+        $contents[] = [
+            'role' => 'user',
+            'parts' => [
+                ['text' => "System instruction: " . $systemPrompt],
+                ['text' => $testMsg]
+            ]
+        ];
+    } else {
+        $contents[] = [
             'role' => 'user',
             'parts' => [['text' => $testMsg]]
-        ]],
-        'systemInstruction' => [
-            'parts' => [['text' => $systemPrompt]]
-        ],
+        ];
+    }
+    
+    $payloadData = [
+        'contents' => $contents,
         'generationConfig' => [
             'responseModalities' => ['AUDIO'],
             'speechConfig' => [
@@ -72,7 +87,15 @@ foreach ($candidateModels as $model) {
                 ]
             ]
         ]
-    ]);
+    ];
+    
+    if ($useSystemInstructionField) {
+        $payloadData['systemInstruction'] = [
+            'parts' => [['text' => $systemPrompt]]
+        ];
+    }
+    
+    $payload = json_encode($payloadData);
 
     $ch = curl_init($url);
     curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
@@ -87,7 +110,8 @@ foreach ($candidateModels as $model) {
     if ($httpCode !== 200) {
         $errorData = json_decode($response, true);
         $errMsg = $errorData['error']['message'] ?? 'Unknown error';
-        echo "❌ Failed (HTTP {$httpCode}: {$errMsg})\n";
+        echo "   ❌ Failed (HTTP {$httpCode}: {$errMsg})\n";
+        echo "   Response: " . substr($response, 0, 500) . "\n\n";
     } else {
         $data = json_decode($response, true);
         $part = $data['candidates'][0]['content']['parts'][0] ?? null;
@@ -97,10 +121,11 @@ foreach ($candidateModels as $model) {
                 'mimeType' => $part['inlineData']['mimeType'] ?? 'audio/mp4'
             ];
             $successfulModel = $model;
-            echo "✅ SUCCESS!\n";
+            echo "   ✅ SUCCESS!\n\n";
             break; // Stop at first successful model
         } else {
-            echo "❌ Failed (No inlineData in response)\n";
+            echo "   ❌ Failed (No inlineData in response)\n";
+            echo "   Response: " . substr($response, 0, 500) . "\n\n";
         }
     }
 }