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

Fix Gemini audio model to gemini-2.0-flash

Browse Source
This commit is contained in:
Hamza-Ayed
2026-05-22 15:15:35 +03:00
parent 2e7cd11f85
commit 891917a1ed
2 changed files with 70 additions and 25 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
backend/app/Services/GeminiService.php
View File

@@ -218,7 +218,7 @@ class GeminiService
*/
public static function generateAudioResponse(string $apiKey, string $systemPrompt, string $userMessage, string $voiceName = 'Puck'): ?array
{
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-flash-lite-latest:generateContent?key=' . $apiKey;
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=' . $apiKey;
$payload = json_encode([
'contents' => [
@@ -280,7 +280,7 @@ class GeminiService
*/
public static function generateAudioResponseFromAudio(string $apiKey, string $systemPrompt, string $audioBase64, string $mimeType, string $voiceName = 'Puck'): ?array
{
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-flash-lite-latest:generateContent?key=' . $apiKey;
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=' . $apiKey;
if (strpos($mimeType, ';') !== false) {
$mimeType = trim(explode(';', $mimeType)[0]);

91
backend/public/test_audio.php
View File

@@ -38,32 +38,77 @@ echo "\n--- Testing Gemini Audio Response ---\n";
$systemPrompt = "You are a friendly customer service assistant. Speak in a warm, welcoming tone.";
$testMsg = "مرحبا، هل متجركم مفتوح اليوم؟";
echo "Sending text prompt: '$testMsg'\n";
$startTime = microtime(true);
$audioResponse = GeminiService::generateAudioResponse($apiKey, $systemPrompt, $testMsg, 'Puck');
$elapsedTime = round(microtime(true) - $startTime, 2);
// Test via cURL directly to dump exact error
$url = 'https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key=' . $apiKey;
$payload = json_encode([
'contents' => [[
'role' => 'user',
'parts' => [['text' => $testMsg]]
]],
'systemInstruction' => [
'parts' => [['text' => $systemPrompt]]
],
'generationConfig' => [
'responseModalities' => ['AUDIO'],
'speechConfig' => [
'voiceConfig' => [
'prebuiltVoiceConfig' => [
'voiceName' => 'Puck'
]
]
]
]
]);
if ($audioResponse && !empty($audioResponse['audio'])) {
echo "✅ [Gemini] Successfully generated voice note in {$elapsedTime} seconds!\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse['mimeType'] . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse['audio']) . " base64 chars\n";
// 3. Test Audio-to-Audio conversion
echo "\n--- Testing Audio-to-Audio (Speech-to-Speech) ---\n";
$startTime = microtime(true);
$audioResponse2 = GeminiService::generateAudioResponseFromAudio($apiKey, $systemPrompt, $audioResponse['audio'], $audioResponse['mimeType'], 'Puck');
$elapsedTime2 = round(microtime(true) - $startTime, 2);
if ($audioResponse2 && !empty($audioResponse2['audio'])) {
echo "✅ [Gemini] Successfully generated Audio-to-Audio response in {$elapsedTime2} seconds!\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse2['mimeType'] . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse2['audio']) . " base64 chars\n";
$ch = curl_init($url);
curl_setopt($ch, CURLOPT_RETURNTRANSFER, true);
curl_setopt($ch, CURLOPT_POST, true);
curl_setopt($ch, CURLOPT_POSTFIELDS, $payload);
curl_setopt($ch, CURLOPT_HTTPHEADER, ['Content-Type: application/json']);
curl_setopt($ch, CURLOPT_TIMEOUT, 30);
$response = curl_exec($ch);
$httpCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
curl_close($ch);
if ($httpCode !== 200) {
echo "❌ [Gemini API Error] HTTP Status: " . $httpCode . "\n";
echo "Response Payload: " . $response . "\n";
$audioResponse = null;
} else {
echo "❌ [Gemini] Audio-to-Audio generation failed. Check errors below:\n";
$data = json_decode($response, true);
$part = $data['candidates'][0]['content']['parts'][0] ?? null;
if ($part && isset($part['inlineData'])) {
$audioResponse = [
'audio' => $part['inlineData']['data'],
'mimeType' => $part['inlineData']['mimeType'] ?? 'audio/mp4'
];
} else {
echo "❌ [Gemini API Error] Response did not contain inlineData. Response: " . $response . "\n";
$audioResponse = null;
}
}
if ($audioResponse && !empty($audioResponse['audio'])) {
echo "✅ [Gemini] Successfully generated voice note!\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse['mimeType'] . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse['audio']) . " base64 chars\n";
// 3. Test Audio-to-Audio conversion
echo "\n--- Testing Audio-to-Audio (Speech-to-Speech) ---\n";
$startTime = microtime(true);
$audioResponse2 = GeminiService::generateAudioResponseFromAudio($apiKey, $systemPrompt, $audioResponse['audio'], $audioResponse['mimeType'], 'Puck');
$elapsedTime2 = round(microtime(true) - $startTime, 2);
if ($audioResponse2 && !empty($audioResponse2['audio'])) {
echo "✅ [Gemini] Successfully generated Audio-to-Audio response in {$elapsedTime2} seconds!\n";
echo "ℹ️ [Gemini] Audio MimeType: " . $audioResponse2['mimeType'] . "\n";
echo "ℹ️ [Gemini] Audio Size: " . strlen($audioResponse2['audio']) . " base64 chars\n";
} else {
echo "❌ [Gemini] Audio-to-Audio generation failed.\n";
}
} else {
echo "❌ [Gemini] Audio response generation failed.\n";
}
} else {
echo "❌ [Gemini] Audio response generation failed. Please check the logs.\n";
}
// 4. Check Node.js WhatsApp Gateway Status
echo "\n--- Checking Node.js Gateway Status ---\n";