Download Turkish female and male speech datasets from Mozilla Common Voice for speech synthesis (TTS) systems

This repository provides Turkish voice datasets of male and female speakers from Mozilla Common Voice, public audio data that you can download for speech synthesis systems.

Source Dataset	Language	Gender	# of Sentences	Size (Hours)
Common Voice Corpus 11.0	Turkish	Female (1 speaker)	9960	9.35
Common Voice Corpus 11.0	Turkish	Male (1 speaker)	4999	3.09

Download Speech Data for Any Languages for Specific Speaker

• In the following script download_tr_female_and_male_speaker_audios.ipynb change the version and language to you wish to download from Mozilla.

version = 'cv-corpus-11.0-2022-09-21'
language = 'cv-corpus-11.0-2022-09-21-tr'

You can list speakers who have the most recorded audios.

Client ID: ba25f19f22267b1f863cef586f17252a9067bc0ecb3d63dd7626fbce7014a16412dd64d5404d34505c0c90635adbc3cb2d830d77291a4db194e5abad91160957 | Gender: female | Count: 17252
Client ID: 6cf4dea5df1dfe5dc5d7bacdb6da105b9e48e01fd0f069dbaf0ad906a7c5d78032194a3a9384e8589b93c58a04a12f676669f4c4501e52cee652f331b72b9205 | Gender: female | Count: 17194
Client ID: c3c204ffaebfc46c0265773376f9288a372694aa79f97afe224828033c28d6d2a90919aaf769bf14105cb4033650e10275760afe250490782eae15c1d1518799 | Gender: male | Count: 6349
Client ID: 60cee2235d7ec4cdeb89d601b8c373955b303c712ec729ed0affdabda8819908f51cefa990163d2bc4ac04c93e6dac2909cca67829211df4a2a17af2507dd50a | Gender: male | Count: 6291
Client ID: ca179eb54e4e584d30587ba3c6c1d4de7d9082d649113d481d308f3e7d858e0a1df21d9591f9c613928b1e122aa7f54535d0f0e0c1380d87d8017844e6e73ea2 | Gender: male | Count: 4837

• Choose the speaker ID that you want to download the dataset.

female_rank = input("Choose the ID number rank for the female voice dataset:")
male_rank = input("Choose the ID number rank for the male voice dataset:")

• Finally you can download ./CommonVoice_dataset/female_data/wav/ and ./CommonVoice_dataset/male_data/wav/ datasets; create ./commonvoice_female_data_manifest.jsonl and ./commonvoice_female_data_manifest.jsonl/ manifest files.

Female voice dataset total duration:9.35 hours
Male voice  dataset total duration:3.09 hours