{ "token2id": { "": 0, "": 1, "": 2, "AA": 3, "AE": 4, "AH": 5, "AO": 6, "AW": 7, "AY": 8, "B": 9, "CH": 10, "D": 11, "DH": 12, "EH": 13, "ER": 14, "EY": 15, "F": 16, "G": 17, "HH": 18, "IH": 19, "IY": 20, "JH": 21, "K": 22, "L": 23, "M": 24, "N": 25, "NG": 26, "OW": 27, "OY": 28, "P": 29, "R": 30, "S": 31, "SH": 32, "T": 33, "TH": 34, "UH": 35, "UW": 36, "V": 37, "W": 38, "Y": 39, "Z": 40, "ZH": 41 }, "speaker_map": { "9017": 0 }, "target_speaker": "9017", "dataset": "MikhailT/hifi-tts", "n_speakers": 1, "phoneme_vocab_size": 42, "sample_rate": 24000, "n_mels": 100, "hop_length": 256, "win_length": 1024, "n_fft": 1024, "f_min": 0.0, "f_max": 12000.0, "min_dur": 1, "pitch_mean": 5.168822016439073, "pitch_std": 0.5352270426450851, "energy_mean": 26.06722209145705, "energy_std": 14.524832135715863, "n_train": 51254, "n_val": 200 }