Most natural language processing (NLP) studies need language-specific data for that language. Some languages like Turkish have scarce data sources to train successful deep learning models. Studies like speech synthesis require dozens of hours of professionally recorded speech with its correct transcription. Creating or finding datasets for text-to-speech (TTS) studies can be quite costly for both time and financial perspectives. This study tries to observe whether English acoustic data can be used to train a Turkish text-to-speech model to eliminate the data problem.