Spaces:
Building
Building
Update stt/stt_google.py
Browse files- stt/stt_google.py +14 -0
stt/stt_google.py
CHANGED
|
@@ -36,20 +36,34 @@ class GoogleSTT(STTInterface):
|
|
| 36 |
"""Map language codes to Google format"""
|
| 37 |
# Google uses BCP-47 language codes
|
| 38 |
language_map = {
|
|
|
|
| 39 |
"tr-TR": "tr-TR",
|
|
|
|
| 40 |
"en-US": "en-US",
|
| 41 |
"en-GB": "en-GB",
|
|
|
|
| 42 |
"de-DE": "de-DE",
|
|
|
|
| 43 |
"fr-FR": "fr-FR",
|
|
|
|
| 44 |
"es-ES": "es-ES",
|
|
|
|
| 45 |
"it-IT": "it-IT",
|
|
|
|
| 46 |
"pt-BR": "pt-BR",
|
|
|
|
| 47 |
"ru-RU": "ru-RU",
|
|
|
|
| 48 |
"ja-JP": "ja-JP",
|
|
|
|
| 49 |
"ko-KR": "ko-KR",
|
|
|
|
| 50 |
"zh-CN": "zh-CN",
|
|
|
|
| 51 |
"ar-SA": "ar-SA",
|
| 52 |
}
|
|
|
|
|
|
|
| 53 |
return language_map.get(language, language)
|
| 54 |
|
| 55 |
async def transcribe(self, audio_data: bytes, config: STTConfig) -> Optional[TranscriptionResult]:
|
|
|
|
| 36 |
"""Map language codes to Google format"""
|
| 37 |
# Google uses BCP-47 language codes
|
| 38 |
language_map = {
|
| 39 |
+
"tr": "tr-TR",
|
| 40 |
"tr-TR": "tr-TR",
|
| 41 |
+
"en": "en-US",
|
| 42 |
"en-US": "en-US",
|
| 43 |
"en-GB": "en-GB",
|
| 44 |
+
"de": "de-DE",
|
| 45 |
"de-DE": "de-DE",
|
| 46 |
+
"fr": "fr-FR",
|
| 47 |
"fr-FR": "fr-FR",
|
| 48 |
+
"es": "es-ES",
|
| 49 |
"es-ES": "es-ES",
|
| 50 |
+
"it": "it-IT",
|
| 51 |
"it-IT": "it-IT",
|
| 52 |
+
"pt": "pt-BR",
|
| 53 |
"pt-BR": "pt-BR",
|
| 54 |
+
"ru": "ru-RU",
|
| 55 |
"ru-RU": "ru-RU",
|
| 56 |
+
"ja": "ja-JP",
|
| 57 |
"ja-JP": "ja-JP",
|
| 58 |
+
"ko": "ko-KR",
|
| 59 |
"ko-KR": "ko-KR",
|
| 60 |
+
"zh": "zh-CN",
|
| 61 |
"zh-CN": "zh-CN",
|
| 62 |
+
"ar": "ar-SA",
|
| 63 |
"ar-SA": "ar-SA",
|
| 64 |
}
|
| 65 |
+
|
| 66 |
+
# Default to the language itself if not in map
|
| 67 |
return language_map.get(language, language)
|
| 68 |
|
| 69 |
async def transcribe(self, audio_data: bytes, config: STTConfig) -> Optional[TranscriptionResult]:
|