{"created_at":"2026-09-15T14:50:59+03:00","status":"completed","instance_id":50735875,"gpu":"NVIDIA A100-SXM4-40GB","dataset":"Seed-TTS-Eval English","samples_per_model":1088,"models":33,"scored_audio":35904,"completed_shards":198,"dataset_sha256":"8a9386efb1768a90ffba66f8931e3887b53e9c9d1bf246eb46a8cce32bc7b1c1","suite_sha256":"1a37627704da3f0158d29a0cc3cff712494dadd0f03553fea2511a01d60ea8a9","source_commit":"3ce0a0aa9cf3c9b24d8e4d6c3d0315b7bf7adc0a","asr":{"checkpoint":"Systran/faster-whisper-large-v3","revision":"edaa852ec7e145841d8ffdb056a99866b5f0a478","device":"cuda","compute_type":"float16","language":"en","beam_size":5,"temperature":0,"condition_on_previous_text":false,"vad_filter":false},"normalization":"whisper_english","seed":42,"repeats":1,"ci_method":"95% prompt-cluster bootstrap; 1000 resamples; RNG seed 42; full 1088 prompts","protocol":"Fixed provider voice/reference intelligibility; not official zero-shot speaker identity/SIM evaluation","backup_verified_at":"2026-09-15T11:48:02.499471+00:00","dataset_id":"kadirnar/voicehub-arena-seed-tts-eval","dataset_revision":"7d02b49322a9731ea9b1cc6279ef3f7a80949ae8","table":[{"model":"kokoro","name":"Kokoro","checkpoint":"hexgrad/Kokoro-82M","revision":"cbc78411372edb46f7e42030b241834a55fe0cb6","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.007682849762477258,0.011640828342247611],"cer_ci95":[0.001896678239861543,0.0029770741117294947],"wer":0.009629071422590639,"cer":0.0024400404689638756,"mer":0.00962423633776885,"wil":0.01742962890275368,"wip":0.9825703710972463,"exact_match_rate":0.9136029411764706,"utterance_mean_wer":0.009572094061983769,"utterance_mean_cer":0.002414540728261279,"rtf":0.018666759712283744,"latency_p50_s":0.08802692033350468,"latency_p95_s":0.10426744297146796,"peak_vram_mib":478.27685546875,"audio_seconds":4623.275,"silence_ratio":0.3701472327051314,"clipping_ratio":0.0,"rms_dbfs":-26.425495242960476},{"model":"supertonic","name":"Supertonic","checkpoint":"Supertone/supertonic-3","revision":"3cadd1ee6394adea1bd021217a0e650ede09a323","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.00785054517897158,0.011805873826184896],"cer_ci95":[0.002118013958236292,0.003225100490647234],"wer":0.009796533534287867,"cer":0.002663214902100815,"mer":0.009789156626506024,"wil":0.017839658400852665,"wip":0.9821603415991473,"exact_match_rate":0.9099264705882353,"utterance_mean_wer":0.009449587931955387,"utterance_mean_cer":0.002529207763588537,"rtf":0.06423751407986668,"latency_p50_s":0.2797527089715004,"latency_p95_s":0.3004460956901312,"peak_vram_mib":419.396484375,"audio_seconds":4756.657505668934,"silence_ratio":0.43568262751000425,"clipping_ratio":0.0,"rms_dbfs":-26.68671776510351},{"model":"omnivoice","name":"OmniVoice","checkpoint":"k2-fsa/OmniVoice","revision":"c5fdb5ccb189668d56333f77ba2629f4cd7535f4","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.007861317788115801,0.012080484304594895],"cer_ci95":[0.0018904921536443545,0.003657155632574282],"wer":0.009880264590136481,"cer":0.002648336606558353,"mer":0.009874476987447698,"wil":0.0173500167374373,"wip":0.9826499832625627,"exact_match_rate":0.9154411764705882,"utterance_mean_wer":0.009585752278766985,"utterance_mean_cer":0.0024681703906616025,"rtf":0.328981732833577,"latency_p50_s":1.332394178956747,"latency_p95_s":1.5358351096510887,"peak_vram_mib":3228.8759765625,"audio_seconds":4436.8,"silence_ratio":0.3348716952743415,"clipping_ratio":0.0,"rms_dbfs":-23.852963743668997},{"model":"speecht5","name":"SpeechT5","checkpoint":"microsoft/speecht5_tts","revision":"30fcde30f19b87502b8435427b5f5068e401d5f6","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.008283238600798748,0.012222576854350246],"cer_ci95":[0.00235856560816167,0.003572732489586244],"wer":0.010131457757682325,"cer":0.0029459025174076057,"mer":0.010123828647925033,"wil":0.01858563704868821,"wip":0.9814143629513118,"exact_match_rate":0.90625,"utterance_mean_wer":0.009766929608381814,"utterance_mean_cer":0.002814133144678633,"rtf":0.1622327560700419,"latency_p50_s":0.6036052946001291,"latency_p95_s":0.9177474867552518,"peak_vram_mib":726.37158203125,"audio_seconds":4101.024,"silence_ratio":0.27561073823840104,"clipping_ratio":0.0,"rms_dbfs":-25.413263645785655},{"model":"f5tts","name":"F5TTS","checkpoint":"F5TTS_v1_Base","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.008544055321581331,0.012617792351167754],"cer_ci95":[0.002511205039574283,0.0039833394587379875],"wer":0.010550113036925395,"cer":0.0031988335416294708,"mer":0.010543933054393306,"wil":0.018842870213572005,"wip":0.981157129786428,"exact_match_rate":0.9090073529411765,"utterance_mean_wer":0.011274173130790778,"utterance_mean_cer":0.004422374102850214,"rtf":0.7815442867831554,"latency_p50_s":2.7469177208840847,"latency_p95_s":2.9705387711524964,"peak_vram_mib":1516.4052734375,"audio_seconds":3739.3813333333333,"silence_ratio":0.21193145480339617,"clipping_ratio":3.4988988015009135e-06,"rms_dbfs":-19.899939920591734},{"model":"styletts2","name":"StyleTTS2","checkpoint":"artifacts/styletts2/epochs_2nd_00020.pth","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.008889246341617306,0.01317792950459316],"cer_ci95":[0.0023622418836487065,0.003664567204361421],"wer":0.010885037260319853,"cer":0.0029905374040349935,"mer":0.010876840696117804,"wil":0.019831987068541723,"wip":0.9801680129314583,"exact_match_rate":0.9044117647058824,"utterance_mean_wer":0.010743798174864353,"utterance_mean_cer":0.0029116890397277357,"rtf":0.04208604152148397,"latency_p50_s":0.19795850105583668,"latency_p95_s":0.2473351741209626,"peak_vram_mib":1068.0185546875,"audio_seconds":5058.533333333333,"silence_ratio":0.29950891529922624,"clipping_ratio":0.0,"rms_dbfs":-22.315291148181117},{"model":"echo","name":"EchoTTS","checkpoint":"jordand/echo-tts-base","revision":"06731f67000241d55cff6b0a961f22f5f151f24f","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.008904209796796562,0.013934323157998304],"cer_ci95":[0.0025651236034547214,0.0047400288829906],"wer":0.01121996148371431,"cer":0.0034815211569362613,"mer":0.011207761793241887,"wil":0.020324327635941386,"wip":0.9796756723640586,"exact_match_rate":0.9034926470588235,"utterance_mean_wer":0.011593386444029826,"utterance_mean_cer":0.003852356493516786,"rtf":0.5723286716880129,"latency_p50_s":2.456841016188264,"latency_p95_s":2.5182850835844874,"peak_vram_mib":9363.14453125,"audio_seconds":4676.359546485261,"silence_ratio":0.12887222618488867,"clipping_ratio":0.0,"rms_dbfs":-13.952829189004829},{"model":"fishtts","name":"FishTTS","checkpoint":"fishaudio/s2-pro","revision":"1de9996b6be38b745688de084d87a5633f714e4e","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.009372852661418718,0.013923650739157823],"cer_ci95":[0.0025164327337067866,0.003856638680410998],"wer":0.011471154651260152,"cer":0.0031690769505445457,"mer":0.011463475859760689,"wil":0.020912083738239273,"wip":0.9790879162617607,"exact_match_rate":0.8970588235294118,"utterance_mean_wer":0.010823033686085157,"utterance_mean_cer":0.002976193604265695,"rtf":3.7434938590228506,"latency_p50_s":14.975895503535867,"latency_p95_s":21.895659835077815,"peak_vram_mib":14652.67431640625,"audio_seconds":4357.735328798186,"silence_ratio":0.13489419262981148,"clipping_ratio":0.0,"rms_dbfs":-16.634614222894925},{"model":"xtts","name":"XTTS","checkpoint":"artifacts/xtts2","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.009479942028329125,0.013929717071918904],"cer_ci95":[0.003047702271545646,0.004894855067600172],"wer":0.011722347818805995,"cer":0.003927870023210141,"mer":0.011700793982448809,"wil":0.020155170204901096,"wip":0.9798448297950989,"exact_match_rate":0.8979779411764706,"utterance_mean_wer":0.012030761823960354,"utterance_mean_cer":0.003932957956043187,"rtf":0.37659089175029953,"latency_p50_s":1.377348417416215,"latency_p95_s":2.1737355709075925,"peak_vram_mib":2044.58837890625,"audio_seconds":4109.9946666666665,"silence_ratio":0.12863922070376485,"clipping_ratio":0.0,"rms_dbfs":-17.494293341348765},{"model":"zonos2","name":"Zonos2","checkpoint":"Zyphra/ZONOS2","revision":"65f1e80f94b599d474bb6af9094a803dc52f60bd","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.009429594420391598,0.01437212173948811],"cer_ci95":[0.0030648629407874664,0.0057465328607648365],"wer":0.01180607887465461,"cer":0.004210557638516931,"mer":0.01179817588486319,"wil":0.020508167521017695,"wip":0.9794918324789823,"exact_match_rate":0.9034926470588235,"utterance_mean_wer":0.011642725636291814,"utterance_mean_cer":0.004352088341262807,"rtf":6.227715330972293,"latency_p50_s":21.817871015518904,"latency_p95_s":32.1434196209535,"peak_vram_mib":15765.91650390625,"audio_seconds":3808.4556916099773,"silence_ratio":0.1599482895773222,"clipping_ratio":0.0,"rms_dbfs":-21.204251726272272},{"model":"chatterbox","name":"Chatterbox","checkpoint":"ResembleAI/chatterbox","revision":"5bb1f6ee58e50c3b8d408bc82a6d3740c2db6e18","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.010239624689752567,0.014502005846153599],"cer_ci95":[0.002878993025473877,0.004195101111888829],"wer":0.012308465209746294,"cer":0.0035261560435636496,"mer":0.012300225922516944,"wil":0.022405080632642016,"wip":0.977594919367358,"exact_match_rate":0.8851102941176471,"utterance_mean_wer":0.011868301735029677,"utterance_mean_cer":0.0034512612340441458,"rtf":0.8215815223732541,"latency_p50_s":2.598566422238946,"latency_p95_s":3.5963106857612717,"peak_vram_mib":3243.5419921875,"audio_seconds":3491.6,"silence_ratio":0.14410049492526528,"clipping_ratio":3.745316243777417e-06,"rms_dbfs":-17.46151259613592},{"model":"mosstts","name":"MossTTS","checkpoint":"OpenMOSS-Team/MOSS-TTS-v1.5","revision":"cdd3b911b1585e3f2dbc7775ef10f9926f58850a","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.009509180623730396,0.015755375124648537],"cer_ci95":[0.002815820019341391,0.006384203524120753],"wer":0.012308465209746294,"cer":0.00431470570731417,"mer":0.012298167823977245,"wil":0.021827881301883956,"wip":0.978172118698116,"exact_match_rate":0.9080882352941176,"utterance_mean_wer":0.012728509235862176,"utterance_mean_cer":0.004697708975017243,"rtf":1.389709319705265,"latency_p50_s":7.407788587734103,"latency_p95_s":10.18555185869335,"peak_vram_mib":25009.28662109375,"audio_seconds":5986.0,"silence_ratio":0.43571122588738487,"clipping_ratio":4.602443648000904e-06,"rms_dbfs":-26.128175142768608},{"model":"outetts","name":"OuteTTS","checkpoint":"OuteAI/Llama-OuteTTS-1.0-1B","revision":"f445d0d301c7fc9e47fd545eb9f9d2912afe7a77","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.010140987840998449,0.014841937026503233],"cer_ci95":[0.003199460828193918,0.0057059813442127735],"wer":0.012392196265594908,"cer":0.00437421888948402,"mer":0.012376651613982272,"wil":0.0210828903062108,"wip":0.9789171096937892,"exact_match_rate":0.8933823529411765,"utterance_mean_wer":0.01187204084722467,"utterance_mean_cer":0.004199895383034018,"rtf":3.164612982009557,"latency_p50_s":14.944048950448632,"latency_p95_s":20.17052680440247,"peak_vram_mib":4373.70263671875,"audio_seconds":5142.874,"silence_ratio":0.3346259142214407,"clipping_ratio":0.0,"rms_dbfs":-19.94094323546973},{"model":"vibevoice","name":"VibeVoice","checkpoint":"microsoft/VibeVoice-Realtime-0.5B","revision":"6bce5f06044837fe6d2c5d7a71a84f0416bd57e4","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.010848131126616388,0.015379696839116497],"cer_ci95":[0.003391400828390384,0.00499105574195605],"wer":0.013145775768232438,"cer":0.004151044456347081,"mer":0.013129285833751463,"wil":0.023229322079901915,"wip":0.9767706779200981,"exact_match_rate":0.8823529411764706,"utterance_mean_wer":0.012995374453257238,"utterance_mean_cer":0.004069196706291303,"rtf":0.6212332774264896,"latency_p50_s":2.793820545077324,"latency_p95_s":4.0911998309195035,"peak_vram_mib":2034.2626953125,"audio_seconds":4941.2,"silence_ratio":0.31937791229579665,"clipping_ratio":7.978451797385622e-09,"rms_dbfs":-26.039530312116856},{"model":"qwen3tts","name":"Qwen3TTS","checkpoint":"Qwen/Qwen3-TTS-12Hz-1.7B-CustomVoice","revision":"0c0e3051f131929182e2c023b9537f8b1c68adfe","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.011347743257994994,0.016629512899164478],"cer_ci95":[0.00341322717010673,0.005637916399275065],"wer":0.013815624215021352,"cer":0.0043890971850264835,"mer":0.013802911159444537,"wil":0.02513313514949267,"wip":0.9748668648505073,"exact_match_rate":0.8832720588235294,"utterance_mean_wer":0.013853437463272021,"utterance_mean_cer":0.004452845678980064,"rtf":1.7696765178685818,"latency_p50_s":8.603621456772089,"latency_p95_s":13.723505410738287,"peak_vram_mib":4351.36962890625,"audio_seconds":5396.08,"silence_ratio":0.2114373210618118,"clipping_ratio":0.0,"rms_dbfs":-21.57469666576863},{"model":"neutts","name":"NeuTTS","checkpoint":"neuphonic/neutts-2e","revision":"412aaab11c6b727c6c0fe2552db305c234e568da","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.011031144307612399,0.019876654385442543],"cer_ci95":[0.003660288780343977,0.010057701099493685],"wer":0.014736665829356108,"cer":0.006204249241206927,"mer":0.014692378328741965,"wil":0.02428765130115995,"wip":0.97571234869884,"exact_match_rate":0.8887867647058824,"utterance_mean_wer":0.014267839703068943,"utterance_mean_cer":0.00575985081832371,"rtf":1.7043660177076607,"latency_p50_s":6.949139470234513,"latency_p95_s":9.845434662140903,"peak_vram_mib":3476.28369140625,"audio_seconds":4472.06,"silence_ratio":0.27913515952061047,"clipping_ratio":4.345555472947041e-06,"rms_dbfs":-19.889727299836053},{"model":"higgstts","name":"HiggsTTS","checkpoint":"bosonai/higgs-tts-2-3b-base","revision":"d80c511612b3040ff2877ce3d408747df1739f11","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.014684263569190567,0.02052255964339718],"cer_ci95":[0.005334938979585008,0.008322318642279253],"wer":0.017416059616511763,"cer":0.006710111289650658,"mer":0.01738113144480655,"wil":0.02976327841893811,"wip":0.9702367215810619,"exact_match_rate":0.859375,"utterance_mean_wer":0.017204365538464803,"utterance_mean_cer":0.006782952438796437,"rtf":1.171588010392337,"latency_p50_s":4.582375276833773,"latency_p95_s":7.186718510463833,"peak_vram_mib":11947.0888671875,"audio_seconds":4356.6,"silence_ratio":0.2434151965391991,"clipping_ratio":2.2705394611234025e-06,"rms_dbfs":-21.519289781106792},{"model":"inflecttts","name":"InflectTTS","checkpoint":"owensong/Inflect-Micro-v2","revision":"132edf8700da8e0d3fa81cdc0fd8844926fc6582","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.014729798756944202,0.020587894869992844],"cer_ci95":[0.004852113065057902,0.006962616006936535],"wer":0.017583521728208994,"cer":0.0058918050348152114,"mer":0.017562933846282512,"wil":0.03222911364088443,"wip":0.9677708863591156,"exact_match_rate":0.8566176470588235,"utterance_mean_wer":0.017602403723727254,"utterance_mean_cer":0.006094378628546624,"rtf":0.014407729416794657,"latency_p50_s":0.060717109590768814,"latency_p95_s":0.07416072003543377,"peak_vram_mib":174.49169921875,"audio_seconds":4074.2906666666668,"silence_ratio":0.23552280891102917,"clipping_ratio":0.0,"rms_dbfs":-23.741855445278556},{"model":"orpheustts","name":"OrpheusTTS","checkpoint":"canopylabs/orpheus-3b-0.1-ft","revision":"4206a56e5a68cf6cf96900a8a78acd3370c02eb6","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.016988210857373835,0.023450519014427323],"cer_ci95":[0.006083833741794046,0.009892766606762127],"wer":0.020011722347818808,"cer":0.007692078795453193,"mer":0.019973257563095436,"wil":0.03503674213176289,"wip":0.9649632578682371,"exact_match_rate":0.8373161764705882,"utterance_mean_wer":0.020058611476350446,"utterance_mean_cer":0.008047379814062864,"rtf":2.5830602675349486,"latency_p50_s":11.744120076298714,"latency_p95_s":17.74872815031558,"peak_vram_mib":6409.50244140625,"audio_seconds":5113.685333333333,"silence_ratio":0.40942533109258156,"clipping_ratio":0.0,"rms_dbfs":-27.90337897777544},{"model":"zonos","name":"Zonos","checkpoint":"Zyphra/Zonos-v0.1-transformer","revision":"9d8331fc49cb5ba8aad2bb56cafd809c66598f4e","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.02007390075144153,0.026153331156134465],"cer_ci95":[0.008174096745526271,0.01130165207207019],"wer":0.02302604035836892,"cer":0.0096411355115158,"mer":0.022951093306626605,"wil":0.037599506467079435,"wip":0.9624004935329206,"exact_match_rate":0.8023897058823529,"utterance_mean_wer":0.023355699436872036,"utterance_mean_cer":0.010020021987216247,"rtf":1.7680858104657158,"latency_p50_s":6.9678456503897905,"latency_p95_s":10.227166134119033,"peak_vram_mib":4144.82080078125,"audio_seconds":4331.996009070294,"silence_ratio":0.26132722105228,"clipping_ratio":1.1080583572800664e-07,"rms_dbfs":-23.673586565392633},{"model":"melotts","name":"MeloTTS","checkpoint":"EN","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.02316930251848474,0.029679677277896818],"cer_ci95":[0.009173332758245023,0.012222008404797419],"wer":0.02637528259231349,"cer":0.01065285960840326,"mer":0.026328986960882646,"wil":0.048229403790975844,"wip":0.9517705962090242,"exact_match_rate":0.7766544117647058,"utterance_mean_wer":0.02676533790963938,"utterance_mean_cer":0.01089498176697487,"rtf":0.019975049763602603,"latency_p50_s":0.1015011165291071,"latency_p95_s":0.11808984335511923,"peak_vram_mib":506.787109375,"audio_seconds":4986.566530612245,"silence_ratio":0.3701696886986826,"clipping_ratio":0.0,"rms_dbfs":-26.93682087807902},{"model":"gptsovits","name":"GPTSoVITS","checkpoint":"lj1995/GPT-SoVITS","revision":"336b2ec4e8d4ac74740798dd40af44e74659ecaf","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.022651802065312383,0.030441862466967787],"cer_ci95":[0.007285552893791007,0.010238613992923822],"wer":0.02637528259231349,"cer":0.008703802892340653,"mer":0.02626532143750521,"wil":0.04524414603704274,"wip":0.9547558539629573,"exact_match_rate":0.8226102941176471,"utterance_mean_wer":0.02771551747884468,"utterance_mean_cer":0.009832378998327021,"rtf":0.3222476706943061,"latency_p50_s":1.066016722470522,"latency_p95_s":1.7044374926015722,"peak_vram_mib":764.4111328125,"audio_seconds":3679.6,"silence_ratio":0.15031361795966175,"clipping_ratio":0.0,"rms_dbfs":-21.53940639823217},{"model":"parlertts","name":"ParlerTTS","checkpoint":"parler-tts/parler-tts-mini-v1","revision":"0392b9451a601e528fd863bbb0598431fee810d9","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.0227290026732326,0.03095461490670336],"cer_ci95":[0.012000777770688552,0.017373967449886223],"wer":0.026542744704010718,"cer":0.0144617032672737,"mer":0.026438698915763136,"wil":0.04502352105370688,"wip":0.9549764789462931,"exact_match_rate":0.8079044117647058,"utterance_mean_wer":0.027006843773840338,"utterance_mean_cer":0.014403632229321195,"rtf":1.8774146532842886,"latency_p50_s":6.555586036294699,"latency_p95_s":11.28402764480561,"peak_vram_mib":4507.4521484375,"audio_seconds":4018.3873015873014,"silence_ratio":0.1515094306108054,"clipping_ratio":7.417982043023822e-09,"rms_dbfs":-18.799491891228392},{"model":"csm","name":"CSM","checkpoint":"sesame/csm-1b","revision":"c92a71e1c419772e25be7dc14d952c2521a740ab","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.02256984690648647,0.03114456456487128],"cer_ci95":[0.012244555826613683,0.017740773268843477],"wer":0.026710206815707946,"cer":0.014878295542462656,"mer":0.026462048942347573,"wil":0.04070971618598751,"wip":0.9592902838140125,"exact_match_rate":0.8088235294117647,"utterance_mean_wer":0.027264451475091614,"utterance_mean_cer":0.015297082046186216,"rtf":2.2531292765703044,"latency_p50_s":8.784898072481155,"latency_p95_s":15.126542358100414,"peak_vram_mib":4330.97998046875,"audio_seconds":4515.44,"silence_ratio":0.3169689076414828,"clipping_ratio":3.525773558828079e-05,"rms_dbfs":-26.242882987863332},{"model":"voxcpm","name":"VoxCPM","checkpoint":"openbmb/VoxCPM2","revision":"bffb3df5a29440629464e5e839f4d214c8714c3d","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.027210668207721767,0.04040970866798127],"cer_ci95":[0.020480165206463754,0.03256538243822571],"wer":0.0334924223394457,"cer":0.02599238231268226,"mer":0.032706459525756335,"wil":0.04076054667222251,"wip":0.9592394533277775,"exact_match_rate":0.8363970588235294,"utterance_mean_wer":0.04057948791404673,"utterance_mean_cer":0.03500154595326482,"rtf":0.8742893887299773,"latency_p50_s":3.0798537097871304,"latency_p95_s":4.482512011565265,"peak_vram_mib":5568.4033203125,"audio_seconds":3915.68,"silence_ratio":0.25207125031657623,"clipping_ratio":0.0,"rms_dbfs":-19.995685779135187},{"model":"openvoice","name":"OpenVoice","checkpoint":"myshell-ai/OpenVoiceV2","revision":"f36e7edfe1684461a8343844af60babc2efbb727","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.03259408128993721,0.0412924532698379],"cer_ci95":[0.013940913785427172,0.017979386205271203],"wer":0.03684166457339027,"cer":0.015904897934892578,"mer":0.03674014696058784,"wil":0.06683946355663117,"wip":0.9331605364433688,"exact_match_rate":0.7205882352941176,"utterance_mean_wer":0.039472592478647844,"utterance_mean_cer":0.01727688301463179,"rtf":0.041566943720588286,"latency_p50_s":0.18259630538523197,"latency_p95_s":0.22698942646384238,"peak_vram_mib":752.9150390625,"audio_seconds":4986.566530612245,"silence_ratio":0.3955355348300485,"clipping_ratio":0.0,"rms_dbfs":-27.237915118797478},{"model":"vits","name":"Vits","checkpoint":"facebook/mms-tts-eng","revision":"c71de0fe7204c83f1c10820a7d696d0b450048ba","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.0568077628669862,0.06791827951441402],"cer_ci95":[0.024251829595567223,0.02939332583322282],"wer":0.062295905551369005,"cer":0.02673629708980539,"mer":0.061881393994843215,"wil":0.10974369514421811,"wip":0.8902563048557819,"exact_match_rate":0.6084558823529411,"utterance_mean_wer":0.0668014646232554,"utterance_mean_cer":0.029005364603199517,"rtf":0.036346979152729886,"latency_p50_s":0.1107128206640482,"latency_p95_s":0.3603286813944578,"peak_vram_mib":161.82177734375,"audio_seconds":4713.184,"silence_ratio":0.24361948208129203,"clipping_ratio":0.0,"rms_dbfs":-18.456575731465676},{"model":"bark","name":"Bark","checkpoint":"suno/bark-small","revision":"1dbd7a128513b8ae4a4e2130fed57b7ac9da5bcd","scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.06542025882580918,0.09126204860193128],"cer_ci95":[0.041101972027671005,0.06016230488018854],"wer":0.07736749560411957,"cer":0.049648872225197885,"mer":0.07522592200602458,"wil":0.11118412348963358,"wip":0.8888158765103664,"exact_match_rate":0.6461397058823529,"utterance_mean_wer":0.08971144156575432,"utterance_mean_cer":0.06200807096575261,"rtf":1.3509386013124312,"latency_p50_s":7.324050920084119,"latency_p95_s":17.736357633210712,"peak_vram_mib":1866.55224609375,"audio_seconds":6735.826666666667,"silence_ratio":0.3688515404183891,"clipping_ratio":4.311951581619864e-05,"rms_dbfs":-25.216558823343817},{"model":"conversationtts","name":"ConversationTTS","checkpoint":"artifacts/conversationtts/model.safetensors","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.06827334380474649,0.10941677412955063],"cer_ci95":[0.05507506031858041,0.11838151593873322],"wer":0.08716402913840743,"cer":0.0824703921918705,"mer":0.08171756024805715,"wil":0.0985030718625245,"wip":0.9014969281374755,"exact_match_rate":0.765625,"utterance_mean_wer":0.09572835930129399,"utterance_mean_cer":0.10447683402668193,"rtf":2.6722347402449587,"latency_p50_s":11.374369565397501,"latency_p95_s":21.11884947624052,"peak_vram_mib":4320.94140625,"audio_seconds":5198.24,"silence_ratio":0.22706196691592928,"clipping_ratio":0.0010200252838765175,"rms_dbfs":-16.481099157390705},{"model":"vui","name":"Vui","checkpoint":"vui-abraham-100m.pt","revision":null,"scored":1088,"quality_review":false,"empty_asr_transcripts":0,"wer_ci95":[0.10961879047597937,0.13331448614544683],"cer_ci95":[0.07456182716011253,0.09409126504207617],"wer":0.12149376203633928,"cer":0.08397310004165923,"mer":0.11439608956165248,"wil":0.15491850701861398,"wip":0.845081492981386,"exact_match_rate":0.4963235294117647,"utterance_mean_wer":0.1461771074776305,"utterance_mean_cer":0.11014174174451084,"rtf":0.4120764092569282,"latency_p50_s":1.9950699619948864,"latency_p95_s":2.9318609584122894,"peak_vram_mib":3020.765625,"audio_seconds":5821.52126984127,"silence_ratio":0.36983417003042174,"clipping_ratio":1.0284511102981117e-05,"rms_dbfs":-16.98248013674346},{"model":"cosyvoice","name":"CosyVoice 3 · 0.5B · 2512","checkpoint":"artifacts/cosyvoice3","revision":null,"scored":1088,"quality_review":true,"empty_asr_transcripts":0,"wer_ci95":[0.12685322593416545,0.14966956384385438],"cer_ci95":[0.07094589764337703,0.08377590503471116],"wer":0.13815624215021352,"cer":0.07709932750104148,"mer":0.13532354629705567,"wil":0.2294798518533584,"wip":0.7705201481466416,"exact_match_rate":0.41911764705882354,"utterance_mean_wer":0.14817157392686525,"utterance_mean_cer":0.08461964252927998,"rtf":0.7114221107347759,"latency_p50_s":2.650021158158779,"latency_p95_s":3.798226011358201,"peak_vram_mib":3431.37841796875,"audio_seconds":4108.52,"silence_ratio":0.13905092033502137,"clipping_ratio":0.0,"rms_dbfs":-2.7491495811524946,"quality_review_reason":"Confirmed HiFT implementation bug in archived audio: magnitude clamp order and final LeakyReLU slope differ from publisher. Historical score invalidated; full corrected rerun started. Eight-text diagnostic is not the full score.","quality_audit":"quality-audit.html","upstream_checkpoint":"FunAudioLLM/Fun-CosyVoice3-0.5B-2512","upstream_revision":"29e01c4e8d000f4bcd70751be16fa94bf3d85a18","version_label":"CosyVoice 3, 0.5B, December 2025 base checkpoint (llm.pt; not RL)","score_status":"invalidated_by_implementation_bug"},{"model":"dia","name":"Dia","checkpoint":"nari-labs/Dia-1.6B-0626","revision":"ef2795fcc29c5abe6ffc91fd33808588b49bbc66","scored":1088,"quality_review":true,"empty_asr_transcripts":183,"wer_ci95":[0.5772596992606132,0.7770414764299733],"cer_ci95":[0.5068358244822976,0.6812881581151855],"wer":0.673532613246253,"cer":0.5928256858894245,"mer":0.5234934270467265,"wil":0.6199658900150475,"wip":0.3800341099849525,"exact_match_rate":0.4522058823529412,"utterance_mean_wer":0.7629098579777105,"utterance_mean_cer":0.6984215456512638,"rtf":3.1889203149742467,"latency_p50_s":35.84547644108534,"latency_p95_s":95.22943165823817,"peak_vram_mib":9100.61865234375,"audio_seconds":15192.758276643992,"silence_ratio":0.19584809755048657,"clipping_ratio":4.931407476340513e-08,"rms_dbfs":-22.07775442813294,"quality_review_reason":"High WER remains under review. Three native/Transformers generation comparisons and independent DAC decoding did not establish a port defect. Empty transcripts and ASR repetition affect scores; diagnostic VAD does not restore missing speech.","quality_audit":"quality-audit.html"},{"model":"llasa","name":"Llasa","checkpoint":"HKUSTAudio/Llasa-1B-Multilingual","revision":"7f094cb62b0a9779b334c60d039a61c5a6e04456","scored":1088,"quality_review":true,"empty_asr_transcripts":16,"wer_ci95":[0.7104861781862367,0.7706671980254735],"cer_ci95":[0.4929818523525938,0.5394510820264022],"wer":0.7399313405342042,"cer":0.514342676902934,"mer":0.6458851045168835,"wil":0.835797418691164,"wip":0.16420258130883603,"exact_match_rate":0.03125,"utterance_mean_wer":0.7637409496456972,"utterance_mean_cer":0.5358505998889113,"rtf":0.8671031698147694,"latency_p50_s":5.18002450093627,"latency_p95_s":8.86913811042904,"peak_vram_mib":5498.3251953125,"audio_seconds":7046.34,"silence_ratio":0.4975273408855499,"clipping_ratio":5.305286515617844e-06,"rms_dbfs":-29.792159225615766,"quality_review_reason":"High WER remains under review. Official top_k=50 differs from native default; setting it alone did not resolve errors. FP32 LM and legacy publisher XCodec2 comparisons agree closely on the audited examples.","quality_audit":"quality-audit.html"}],"audio_bytes":19539199408,"source_url":"https://github.com/BytedanceSpeech/seed-tts-eval","audio_format":"indexed-wav-tar","quality_audit":{"date":"2026-09-15","report":"quality-audit.html","original_scores_preserved":true,"cosyvoice_corrected_full_run":"in_progress","corrected_full_score_available":false}}
