Omarrran's picture
TTS Dataset Collector for HF Spaces
88b6846
raw
history blame contribute delete
854 Bytes
import { franc } from 'franc';
export const RTL_LANGUAGES = ['arb', 'heb', 'urd', 'per', 'ara', 'fas', 'urd'];
export function detectLanguage(text: string): string {
// franc returns 'und' if undetermined
// We can set a minimum length threshold to avoid noise
if (!text || text.length < 5) return 'eng';
return franc(text);
}
export function isRTL(langCode: string): boolean {
return RTL_LANGUAGES.includes(langCode);
}
export function getSuggestedFont(langCode: string): string {
switch (langCode) {
case 'arb':
case 'ara':
return 'Amiri'; // Assuming we have this or similar
case 'jpn':
return 'Noto Sans JP';
case 'kor':
return 'Noto Sans KR';
case 'cmn':
return 'Noto Sans SC';
default:
return 'DM Sans';
}
}