Arabic Text-to-Speech (TTS) system using FastSpeech2, FastPitch, MixerTTS for mel-spectrogram generation, HiFi-GAN as vocoder, and SparkTTS as a fully end-to-end model. Trained on CLArTTS and Arabic Speech Corpus datasets.
text-to-speech deep-learning speech-synthesis fastspeech2 hifi-gan fastpitch mixer-tts spark-tts clar-tts
-
Updated
Dec 17, 2025 - Jupyter Notebook