F5-TTS: A Fairytaler that Fakes Fluent and Faithful Speech with Flow Matching

Benchmark Model Rank Results
text-to-speech-synthesis-on-librispeech-pc-test-cleanF5-TTS (32 NFE)#9WER: 2.42UTMOS: 3.9Speaker Similarity: 0.66
text-to-speech-synthesis-on-seed-tts-test-enF5-TTS (32 NFE)#10WER: 1.83Speaker Similarity: 0.67
text-to-speech-synthesis-on-seed-tts-test-zhF5-TTS (32 NFE)#17CER: 1.56Speaker Similarity: 0.76