Text-to-Speech on Seed-TTS en (test)
0.8WERFish Speech
Evaluation Results
| Method | Links | |||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fish SpeechType=TTS2026.03 | 0.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| WAND (on IndexTTS 1.5)Attn. Mechanism=WAND (W = 32), Cache=O(1), Compute=O(1)2026.03 | 0.91 | — | — | — | — | 4.13 | 3.98 | 0.944 | — | 13.01 | 66.2 | 3.28 | 1.89 | — | — | — | — | |
| IndexTTS 1.5Attn. Mechanism=MHA (W = ∞), Cache=O(L), Compute=O(L)2026.03 | 0.98 | — | — | — | — | 4.13 | 3.97 | 0.944 | — | 38.44 | — | 6.18 | — | — | — | — | — | |
| Qwen3-TTS-12Hz-1.7B-Basezero-shot=true, frame_rate=12Hz, parameters=1.7B2026.01 | 1.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTS#Data(h)=100k Multi., #Params=336M2026.03 | 1.26 | — | — | — | — | — | 3.69 | — | 0.67 | — | — | — | — | — | — | — | — | |
| Qwen3.5-Omni-PlusSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.26 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-TTS-12Hz-0.6B-Basezero-shot=true, frame_rate=12Hz, parameters=0.6B2026.01 | 1.32 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoXtream2#Data(h)=40k EN, #Params=462M2026.03 | 1.32 | — | — | — | — | — | 4.05 | — | 0.656 | — | — | — | — | — | — | — | — | |
| Kyutai-TTS#Data(h)=88k EN, #Params=750M2026.03 | 1.34 | — | — | — | — | — | 3.63 | — | 0.689 | — | — | — | — | — | — | — | — | |
| Qwen3-Omni2026.02 | 1.39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-Omni-30B-A3BSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.39 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxCPM2 + SR-FDSteps=42026.07 | 1.41 | 0.76 | — | — | — | — | 3.76 | — | — | — | — | — | — | — | — | 3.07 | 3.65 | |
| GroundtruthAttn. Mechanism=−, Cache=−, Compute=−2026.03 | 1.44 | — | — | — | — | 3.79 | 3.53 | 1 | — | — | — | — | — | — | — | — | — | |
| CosyVoice 32026.02 | 1.45 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice 3zero-shot=true2026.01 | 1.45 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoXtream2-B#Data(h)=40k EN, #Params=462M2026.03 | 1.45 | — | — | — | — | — | 3.56 | — | 0.679 | — | — | — | — | — | — | — | — | |
| CosyVoice 3Setting=Zero-shot, Aspect=Content Consistency2026.04 | 1.45 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GroundtruthHybrid=–, Source=–, Params=–, Layers=–, Data (kh)=–, Ratio=–, Strategy=–, TFLOPs=–2026.03 | 1.47 | — | — | — | — | 3.93 | 3.52 | 1 | — | — | — | — | — | — | — | — | — | |
| ARCHI-TTSSteps=322026.07 | 1.47 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-TTS-25Hz-1.7B-Basezero-shot=true, frame_rate=25Hz, parameters=1.7B2026.01 | 1.49 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Human#Data(h)=-, #Params=-2026.03 | 1.5 | — | — | — | — | — | 3.53 | — | 0.734 | — | — | — | — | — | — | — | — | |
| WavTTSParams=673M, Data (hrs)=100K Emilia, Protocol=Zero-shot2026.06 | 1.5 | 0.65 | — | — | — | — | 3.92 | — | — | — | — | — | — | — | — | — | — | |
| ERNIE 5.02026.02 | 1.54 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-TTSParams.=1.1B, Training Data (hours)=5000k Multi.2026.04 | 1.54 | 0.708 | — | — | — | — | 4.16 | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-TTSZero-shot=true, #Params=1.1B2026.05 | 1.54 | 0.708 | — | — | — | — | 4.16 | — | — | — | — | — | — | — | — | — | — | |
| OmniVoiceParams.=0.8B, Training Data (hours)=581k Multi.2026.04 | 1.6 | 0.741 | — | — | — | — | 3.91 | — | — | — | — | — | — | — | — | — | — | |
| OmniVoiceZero-shot=true, #Params=0.8B2026.05 | 1.6 | 0.741 | — | — | — | — | 3.91 | — | — | — | — | — | — | — | — | — | — | |
| ZipVoiceParams=123M, Data (hrs)=100K Emilia, Protocol=Zero-shot2026.06 | 1.6 | 0.7 | — | — | — | — | 3.83 | — | — | — | — | — | — | — | — | — | — | |
| Qwen3-TTS-25Hz-0.6B-Basezero-shot=true, frame_rate=25Hz, parameters=0.6B2026.01 | 1.64 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MiniMax-Speechzero-shot=true2026.01 | 1.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| MiniMax-SpeechSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.65 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSParams=336M, Data (hrs)=100K Emilia, Protocol=Zero-shot2026.06 | 1.65 | 0.66 | — | — | — | — | 3.73 | — | — | — | — | — | — | — | — | — | — | |
| ZipVoiceParams.=0.1B, Training Data (hours)=100k Emilia2026.04 | 1.7 | 0.697 | — | — | — | — | 3.82 | — | — | — | — | — | — | — | — | — | — | |
| ZipVoiceZero-shot=true, #Params=0.1B2026.05 | 1.7 | 0.697 | — | — | — | — | 3.82 | — | — | — | — | — | — | — | — | — | — | |
| OmniVoice-EmiliaParams.=0.8B, Training Data (hours)=100k Emilia2026.04 | 1.72 | 0.717 | — | — | — | — | 3.88 | — | — | — | — | — | — | — | — | — | — | |
| WAND (on CosyVoice2-0.5B)Attn. Mechanism=WAND (W = 32), Cache=O(1), Compute=O(1)2026.03 | 1.72 | — | — | — | — | 4.02 | 4.21 | 0.945 | — | 5.25 | 49.9 | 7.44 | 1.55 | — | — | — | — | |
| OmniVoice-EmiliaZero-shot=true, #Params=0.8B2026.05 | 1.72 | 0.717 | — | — | — | — | 3.88 | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTS DiT2024.10 | 1.733 | 0.79 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxCPM2Steps=102026.07 | 1.74 | 0.76 | — | — | — | — | 3.81 | — | — | — | — | — | — | — | — | 3.09 | 3.67 | |
| DELTA-TTSType=Ours, Params.=0.5B+94M, Training Data (hours)=0.585K LibriTTS2026.07 | 1.75 | 0.688 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Ground TruthParams=–, Data (hrs)=–, Protocol=Zero-shot2026.06 | 1.79 | 0.73 | — | — | — | — | 3.53 | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSType=TTS2026.03 | 1.8 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSNFE=322024.10 | 1.83 | 0.67 | 0.31 | 3.89 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTS2026.02 | 1.83 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSzero-shot=true2026.01 | 1.83 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.83 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTS2026.06 | 1.83 | 0.647 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSSteps=322026.07 | 1.83 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark-TTS#Data(h)=102k Multi., #Params=507M2026.03 | 1.85 | — | — | — | — | — | 3.94 | — | 0.572 | — | — | — | — | — | — | — | — | |
| F5-TTSParams.=0.4B, Training Data (hours)=100K Emilia2026.04 | 1.85 | 0.664 | — | — | — | — | 3.72 | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSZero-shot=true, #Params=0.4B2026.05 | 1.85 | 0.664 | — | — | — | — | 3.72 | — | — | — | — | — | — | — | — | — | — | |
| VoxCPMType=AR, Params.=0.5B, Training Data (hours)=1800K Multi.2026.07 | 1.85 | 0.729 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSNFE=162024.10 | 1.89 | 0.67 | 0.16 | 3.79 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxCPMParams.=0.7B, Training Data (hours)=1800k Multi.2026.04 | 1.92 | 0.731 | — | — | — | — | 3.77 | — | — | — | — | — | — | — | — | — | — | |
| VoxCPMZero-shot=true, #Params=0.7B2026.05 | 1.92 | 0.731 | — | — | — | — | 3.77 | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTS_RL2026.02 | 1.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| KALL-Ezero-shot=true2026.01 | 1.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice2-0.5BAttn. Mechanism=GQA (W = ∞), Cache=O(L), Compute=O(L)2026.03 | 1.94 | — | — | — | — | 4.01 | 4.18 | 0.946 | — | 10.48 | — | 11.55 | — | — | — | — | — | |
| Seed-TTS RLSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.94 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| LongCat-AudioDiTParams=1420M, Data (hrs)=100K Multi., Protocol=Zero-shot2026.06 | 1.94 | 0.76 | — | — | — | — | 3.8 | — | — | — | — | — | — | — | — | — | — | |
| FireRedTTS 2zero-shot=true2026.01 | 1.95 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| FireRedTTS2Type=AR, Params.=1.5B, Training Data (hours)=1400K Multi.2026.07 | 1.95 | 0.665 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Chatterbox-FlashZero-shot=true, Decoding method=PMI (α = 0), #Params=0.5B, Steps=82026.05 | 1.96 | 0.704 | — | — | — | — | 4.09 | — | — | — | — | — | — | — | — | — | — | |
| Chatterbox-FlashZero-shot=true, Decoding method=TS schedule, #Params=0.5B, Steps=82026.05 | 1.97 | 0.703 | — | — | — | — | 4.09 | — | — | — | — | — | — | — | — | — | — | |
| Spark TTS2026.02 | 1.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark TTSAccess=Open-Source2026.02 | 1.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark TTSzero-shot=true2026.01 | 1.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark TTSSetting=Zero-shot, Aspect=Content Consistency2026.04 | 1.98 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark-TTSParams=507M, Data (hrs)=102K Multi., Protocol=Zero-shot2026.06 | 1.98 | 0.58 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark-TTS2026.06 | 1.98 | 0.584 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Spark TTSType=AR, Params.=0.5B, Training Data (hours)=100K Multi.2026.07 | 1.98 | 0.573 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| F5-TTSType=NAR, Params.=0.3B, Training Data (hours)=100K Emilia, NFE=322026.07 | 2 | 0.647 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice3Type=AR, Params.=0.5B, Training Data (hours)=1000K Multi.2026.07 | 2.02 | 0.692 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice 2Hybrid=×, Source=Public, Params=0.5B, Layers=24, Data (kh)=170, Ratio=1:0, Strategy=–, TFLOPs=1.782026.03 | 2.03 | — | — | — | — | 3.68 | 4.15 | 0.66 | — | — | — | — | — | — | — | — | — | |
| MaskGCT#Data(h)=100k Multi., #Params=1048M2026.03 | 2.03 | — | — | — | — | — | 3.56 | — | 0.712 | — | — | — | — | — | — | — | — | |
| Chatterbox-FlashZero-shot=true, Decoding method=PMI + ED (α = 0.5), #Params=0.5B, Steps=6.42026.05 | 2.04 | 0.704 | — | — | — | — | 4.08 | — | — | — | — | — | — | — | — | — | — | |
| Ground Truth2024.10 | 2.06 | 0.73 | 0 | 3.91 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Vocoder Resynthesized2024.10 | 2.09 | 0.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Human2026.02 | 2.14 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Ground-truth2026.04 | 2.14 | 0.734 | — | — | — | — | 3.52 | — | — | — | — | — | — | — | — | — | — | |
| Ground-truthZero-shot=true2026.05 | 2.14 | 0.734 | — | — | — | — | 3.52 | — | — | — | — | — | — | — | — | — | — | |
| Ground-truthType=–2026.07 | 2.14 | 0.734 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice3Params.=1.1B, Training Data (hours)=1000k Multi.2026.04 | 2.17 | 0.696 | — | — | — | — | 3.96 | — | — | — | — | — | — | — | — | — | — | |
| CosyVoice3Zero-shot=true, #Params=1.1B2026.05 | 2.17 | 0.696 | — | — | — | — | 3.96 | — | — | — | — | — | — | — | — | — | — | |
| E2 TTSNFE=322024.10 | 2.19 | 0.71 | 0.06 | 3.81 | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| E2 TTS2026.02 | 2.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| E2 TTSzero-shot=true2026.01 | 2.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| E2 TTSSetting=Zero-shot, Aspect=Content Consistency2026.04 | 2.19 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| E2 TTSType=NAR, Params.=0.3B, Training Data (hours)=100K Emilia, NFE=322026.07 | 2.19 | 0.71 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoiceStar#Data(h)=65k EN, #Params=840M2026.03 | 2.2 | — | — | — | — | — | 3.92 | — | 0.605 | — | — | — | — | — | — | — | — | |
| ChatterboxZero-shot=true, #Params=0.5B2026.05 | 2.2 | 0.685 | — | — | — | — | 4.1 | — | — | — | — | — | — | — | — | — | — | |
| E2-TTSParams=333M, Data (hrs)=100K Emilia, Protocol=Zero-shot2026.06 | 2.21 | 0.71 | — | — | — | — | 3.2 | — | — | — | — | — | — | — | — | — | — | |
| FlashTTSStage=12026.06 | 2.21 | 0.572 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| IndexTTS2Type=AR, Params.=1.5B, Training Data (hours)=55K Emilia2026.07 | 2.23 | 0.706 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| VoxCPM2Steps=42026.07 | 2.23 | 0.74 | — | — | — | — | 3.3 | — | — | — | — | — | — | — | — | 2.9 | 3.53 | |
| Seed-TTS_ICL2026.02 | 2.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTS ICLSetting=Zero-shot, Aspect=Content Consistency2026.04 | 2.24 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTSAccess=Proprietary2026.02 | 2.25 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTSzero-shot=true2026.01 | 2.25 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Seed-TTS2026.06 | 2.25 | 0.762 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — |