Hamza/nabeh
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update test_audio to try candidate models

Browse Source
This commit is contained in:
Hamza-Ayed
2026-05-22 15:22:58 +03:00
parent 03a36379ed
commit f339488517
1 changed files with 95 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

120
backend/public/test_audio.php
View File

@@ -33,13 +33,27 @@ if (empty($apiKey)) {
echo "✅ [Database] Found active Gemini rule for Company ID: {$rule['company_id']}\n"; echo "✅ [Database] Found active Gemini rule for Company ID: {$rule['company_id']}\n";
echo "ℹ️ [Config] Gemini API Key starts with: " . substr($apiKey, 0, 6) . "...\n"; echo "ℹ️ [Config] Gemini API Key starts with: " . substr($apiKey, 0, 6) . "...\n";
// 2. Test Gemini API Audio Output (Puck Voice, gemini-flash-lite-latest) // 2. Test Gemini API Audio Output (Puck Voice, multiple models)
echo "\n--- Testing Gemini Audio Response ---\n"; echo "\n--- Testing Gemini Audio Response ---\n";
$systemPrompt = "You are a friendly customer service assistant. Speak in a warm, welcoming tone."; $systemPrompt = "You are a friendly customer service assistant. Speak in a warm, welcoming tone.";
$testMsg = "مرحبا، هل متجركم مفتوح اليوم؟"; $testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";
// Test via cURL directly to dump exact error $candidateModels = [
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=' . $apiKey; 'gemini-2.5-flash-preview-tts',
'gemini-3.1-flash-tts-preview',
'gemini-2.0-flash',
'gemini-2.5-flash',
'gemini-3.5-flash',
'gemini-flash-latest'
];
$successfulModel = null;
$audioResponse = null;
foreach ($candidateModels as $model) {
echo "Testing model: models/{$model} ... ";
$url = 'https://generativelanguage.googleapis.com/v1beta/models/' . $model . ':generateContent?key=' . $apiKey;
$payload = json_encode([ $payload = json_encode([
'contents' => [[ 'contents' => [[
'role' => 'user', 'role' => 'user',
@@ -65,15 +79,15 @@ $testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";
curl_setopt($ch, CURLOPT_POST, true); curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_POSTFIELDS, $payload); curl_setopt($ch, CURLOPT_POSTFIELDS, $payload);
curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']); curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']);
curl_setopt($ch, CURLOPT_TIMEOUT, 30); curl_setopt($ch, CURLOPT_TIMEOUT, 15);
$response = curl_exec($ch); $response = curl_exec($ch);
$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE); $httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
curl_close($ch); curl_close($ch);
if ($httpCode !== 200) { if ($httpCode !== 200) {
echo "❌ [Gemini API Error] HTTP Status: " . $httpCode . "\n"; $errorData = json_decode($response, true);
echo "Response Payload: " . $response . "\n"; $errMsg = $errorData['error']['message'] ?? 'Unknown error';
$audioResponse = null; echo "❌ Failed (HTTP {$httpCode}: {$errMsg})\n";
} else { } else {
$data = json_decode($response, true); $data = json_decode($response, true);
$part = $data['candidates'][0]['content']['parts'][0] ?? null; $part = $data['candidates'][0]['content']['parts'][0] ?? null;
@@ -82,33 +96,89 @@ $testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";
'audio' => $part['inlineData']['data'], 'audio' => $part['inlineData']['data'],
'mimeType' => $part['inlineData']['mimeType'] ?? 'audio/mp4' 'mimeType' => $part['inlineData']['mimeType'] ?? 'audio/mp4'
]; ];
$successfulModel = $model;
echo "✅ SUCCESS!\n";
break; // Stop at first successful model
} else { } else {
echo "❌ [Gemini API Error] Response did not contain inlineData. Response: " . $response . "\n"; echo "❌ Failed (No inlineData in response)\n";
$audioResponse = null;
} }
} }
}
if ($audioResponse && !empty($audioResponse['audio'])) { if ($successfulModel && $audioResponse && !empty($audioResponse['audio'])) {
echo "✅ [Gemini] Successfully generated voice note!\n"; echo "\n✅ [Gemini] Successfully generated voice note using model: models/{$successfulModel}\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse['mimeType'] . "\n"; echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse['mimeType'] . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse['audio']) . " base64 chars\n"; echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse['audio']) . " base64 chars\n";
// 3. Test Audio-to-Audio conversion // 3. Test Audio-to-Audio conversion using the successful model
echo "\n--- Testing Audio-to-Audio (Speech-to-Speech) ---\n"; echo "\n--- Testing Audio-to-Audio (Speech-to-Speech) ---\n";
$startTime = microtime(true); $startTime = microtime(true);
$audioResponse2 = GeminiService::generateAudioResponseFromAudio($apiKey, $systemPrompt, $audioResponse['audio'], $audioResponse['mimeType'], 'Puck');
$elapsedTime2 = round(microtime(true) - $startTime, 2); // We temporarily override the model inside GeminiService for this test, but since GeminiService is not updated yet,
// we'll run a direct curl call for testing audio-to-audio:
if ($audioResponse2 && !empty($audioResponse2['audio'])) { $url = 'https://generativelanguage.googleapis.com/v1beta/models/' . $successfulModel . ':generateContent?key=' . $apiKey;
$payload2 = json_encode([
'contents' => [
[
'role' => 'user',
'parts' => [
[
'inlineData' => [
'mimeType' => $audioResponse['mimeType'],
'data' => $audioResponse['audio']
]
],
[
'text' => "استمع إلى التسجيل الصوتي المرفق وأجب عليه مباشرة بصوتك بناءً على الإرشادات المحددة."
]
]
]
],
'systemInstruction' => [
'parts' => [['text' => $systemPrompt]]
],
'generationConfig' => [
'responseModalities' => ['AUDIO'],
'speechConfig' => [
'voiceConfig' => [
'prebuiltVoiceConfig' => [
'voiceName' => 'Puck'
]
]
]
]
]);
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_POSTFIELDS, $payload2);
curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']);
curl_setopt($ch, CURLOPT_TIMEOUT, 30);
$response2 = curl_exec($ch);
$httpCode2 = curl_getinfo($ch, CURLINFO_HTTP_CODE);
curl_close($ch);
$elapsedTime2 = round(microtime(true) - $startTime, 2);
if ($httpCode2 === 200) {
$data2 = json_decode($response2, true);
$part2 = $data2['candidates'][0]['content']['parts'][0] ?? null;
if ($part2 && isset($part2['inlineData'])) {
echo "✅ [Gemini] Successfully generated Audio-to-Audio response in {$elapsedTime2} seconds!\n"; echo "✅ [Gemini] Successfully generated Audio-to-Audio response in {$elapsedTime2} seconds!\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse2['mimeType'] . "\n"; echo "ℹ️ [Gemini] Audio MimeType: " . ($part2['inlineData']['mimeType'] ?? 'audio/mp4') . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse2['audio']) . " base64 chars\n"; echo "ℹ️ [Gemini] Audio Size: " . strlen($part2['inlineData']['data']) . " base64 chars\n";
} else { } else {
echo "❌ [Gemini] Audio-to-Audio generation failed.\n"; echo "❌ [Gemini] Audio-to-Audio response did not contain inlineData.\n";
} }
} else { } else {
echo "❌ [Gemini] Audio response generation failed.\n"; $errorData2 = json_decode($response2, true);
$errMsg2 = $errorData2['error']['message'] ?? 'Unknown error';
echo "❌ [Gemini] Audio-to-Audio generation failed. HTTP {$httpCode2}: {$errMsg2}\n";
} }
} else {
echo "\n❌ [Gemini] All model trials for audio response generation failed.\n";
}
// 4. Check Node.js WhatsApp Gateway Status // 4. Check Node.js WhatsApp Gateway Status
echo "\n--- Checking Node.js Gateway Status ---\n"; echo "\n--- Checking Node.js Gateway Status ---\n";