From 1ca1cc89dbf222e4f6ac46f70eae2a174667f30a Mon Sep 17 00:00:00 2001 From: Warren B Date: Thu, 3 Sep 2026 02:42:55 +0100 Subject: [PATCH] webui: expose the quality-critical options existing groups were missing Twenty groups already existed but left out options their families read, so the controls that decide output quality were reachable only through the JSON fallback box. - omnivoice: t_shift, class_temperature, position_temperature, layer_penalty_factor, denoise, preprocess_prompt, postprocess_output and the audio/text chunking knobs (session.cpp:100-116, loader.cpp:34). - irodori_tts: its seven guidance knobs plus duration bounds, no_ref, trim_tail and the instruction field (session.cpp:451-454 and the guidance block). - index_tts2: the whole sampling block -- do_sample, temperature, top_p, top_k, repetition_penalty, num_beams, length_penalty. - chatterbox: min_p, its actual truncation filter, alongside top_p, whose default is a documented no-op, and s3gen_cfg_rate. - qwen3_tts: the sub-talker block and x_vector_only_mode. - minimax_music3: ensemble takes and the flow uncond/hop controls. - seed_vc: f0_condition, auto_f0_adjust and semitone_shift, without which the SVC entry cannot do what SVC is for. - voxcpm1/voxcpm2: retry_badcase controls; heartmula: chunking; neutts, vibevoice, stable_audio, ace_step, minimax_h3 and the four firered-audio groups: the sampling or negative-prompt options each reads. Every name was checked against the family's sources. Two controls that the original change had for omnivoice -- reference_max_seconds and reference_pad_ms -- are left out: nothing in the tree reads them. Validation: python3 -m json.tool webui/configs/model_params.json python3 tools/check_loader_catalog_sync.py # in sync --- webui/configs/model_params.json | 104 ++++++++++++++++++++++++++------ 1 file changed, 84 insertions(+), 20 deletions(-) diff --git a/webui/configs/model_params.json b/webui/configs/model_params.json index 7b3d71bbc..dfbb1251b 100644 --- a/webui/configs/model_params.json +++ b/webui/configs/model_params.json @@ -17,7 +17,12 @@ {"name": "repetition_penalty", "type": "slider", "label": "repetition_penalty", "default": 1.05, "minimum": 1.0, "maximum": 2.0, "step": 0.01}, {"name": "do_sample", "type": "bool", "label": "do_sample", "default": true}, {"name": "instruct", "type": "text", "label": "instruct（仅 VoiceDesign/CustomVoice）", "default": "", "placeholder": "风格/音色指令，Base 版忽略"}, - {"name": "speaker", "type": "text", "label": "speaker（仅 CustomVoice）", "default": "", "placeholder": "内置音色名，其它版忽略"} + {"name": "speaker", "type": "text", "label": "speaker（仅 CustomVoice）", "default": "", "placeholder": "内置音色名，其它版忽略"}, + {"name": "subtalker_do_sample", "type": "bool", "label": "subtalker_do_sample", "label_en": "Sub-talker: sample", "default": true}, + {"name": "subtalker_temperature", "type": "slider", "label": "subtalker_temperature", "label_en": "Sub-talker: temperature", "default": 0.9, "minimum": 0.05, "maximum": 2.0, "step": 0.05, "info_en": "Must be above 0 while subtalker_do_sample is on."}, + {"name": "subtalker_top_k", "type": "number", "label": "subtalker_top_k", "label_en": "Sub-talker: top-k", "default": 50, "minimum": 0, "step": 1, "precision": 0}, + {"name": "subtalker_top_p", "type": "slider", "label": "subtalker_top_p", "label_en": "Sub-talker: top-p", "default": 1.0, "minimum": 0.0, "maximum": 1.0, "step": 0.01}, + {"name": "x_vector_only_mode", "type": "bool", "label": "x_vector_only_mode（仅 VoiceClone）", "label_en": "x_vector_only_mode (VoiceClone only)", "default": false, "info_en": "Condition on the speaker embedding alone instead of in-context reference audio. Ignored by other variants."} ], "vibevoice": [ @@ -27,7 +32,8 @@ {"name": "temperature", "type": "slider", "label": "temperature", "default": 1.0, "minimum": 0.05, "maximum": 2.0, "step": 0.05}, {"name": "top_p", "type": "slider", "label": "top_p", "default": 1.0, "minimum": 0.05, "maximum": 1.0, "step": 0.01}, {"name": "do_sample", "type": "bool", "label": "do_sample", "default": false}, - {"name": "voice_samples", "type": "text", "label": "voice_samples（多说话人，逗号分隔 wav，≤4）", "default": "", "placeholder": "D:/a.wav,D:/b.wav — 用此项时勿再上传参考音色"} + {"name": "voice_samples", "type": "text", "label": "voice_samples（多说话人，逗号分隔 wav，≤4）", "default": "", "placeholder": "D:/a.wav,D:/b.wav — 用此项时勿再上传参考音色"}, + {"name": "top_k", "type": "number", "label": "top_k", "label_en": "Top-k", "default": 50, "minimum": 0, "step": 1, "precision": 0, "info_en": "Must be non-negative; 0 disables top-k."} ], "voxcpm2": [ @@ -35,14 +41,19 @@ {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "text_chunk_mode", "type": "choice", "label": "text_chunk_mode", "default": "tag_aware", "choices": ["default", "tag_aware", "japanese", "endline"]}, {"name": "min_tokens", "type": "number", "label": "min_tokens", "default": 2, "minimum": 0, "step": 1, "precision": 0}, - {"name": "retry_badcase", "type": "bool", "label": "retry_badcase（自动重试异常输出）", "default": true} + {"name": "retry_badcase", "type": "bool", "label": "retry_badcase（自动重试异常输出）", "default": true}, + {"name": "retry_badcase_max_times", "type": "number", "label": "retry_badcase_max_times（最大重试次数）", "label_en": "Retry attempts", "default": 3, "minimum": 1, "step": 1, "precision": 0, "info_en": "Must be positive."}, + {"name": "retry_badcase_ratio_threshold", "type": "number", "label": "retry_badcase_ratio_threshold（时长/字数比阈值）", "label_en": "Retry length-ratio threshold", "default": 6.0, "minimum": 0.1, "step": 0.5, "info_en": "Audio-seconds-per-character ratio above which a take is treated as a bad case. Must be positive."} ], "voxcpm1": [ {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 10, "minimum": 1, "step": 1, "precision": 0, "info": "CFM/DiT 步数"}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "min_tokens", "type": "number", "label": "min_tokens", "default": 2, "minimum": 0, "step": 1, "precision": 0}, - {"name": "retry_badcase", "type": "bool", "label": "retry_badcase（自动重试异常输出）", "default": true} + {"name": "retry_badcase", "type": "bool", "label": "retry_badcase（自动重试异常输出）", "default": true}, + {"name": "text_chunk_mode", "type": "choice", "label": "text_chunk_mode", "label_en": "Text chunk mode", "default": "tag_aware", "choices": ["default", "tag_aware", "japanese", "endline"]}, + {"name": "retry_badcase_max_times", "type": "number", "label": "retry_badcase_max_times（最大重试次数）", "label_en": "Retry attempts", "default": 3, "minimum": 1, "step": 1, "precision": 0, "info_en": "Must be positive."}, + {"name": "retry_badcase_ratio_threshold", "type": "number", "label": "retry_badcase_ratio_threshold（时长/字数比阈值）", "label_en": "Retry length-ratio threshold", "default": 6.0, "minimum": 0.1, "step": 0.5, "info_en": "Audio-seconds-per-character ratio above which a take is treated as a bad case. Must be positive."} ], "miotts": [ @@ -57,7 +68,10 @@ {"name": "exaggeration", "type": "slider", "label": "exaggeration", "default": 0.5, "minimum": 0.0, "maximum": 2.0, "step": 0.05, "info": "改动后需重新『加载模型』才生效"}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 0.5, "minimum": 0.0, "maximum": 1.0, "step": 0.05, "info": "改动后需重新『加载模型』才生效"}, {"name": "temperature", "type": "slider", "label": "temperature", "default": 0.8, "minimum": 0.0, "maximum": 2.0, "step": 0.05}, - {"name": "repetition_penalty", "type": "slider", "label": "repetition_penalty", "default": 1.2, "minimum": 1.0, "maximum": 2.0, "step": 0.01} + {"name": "repetition_penalty", "type": "slider", "label": "repetition_penalty", "default": 1.2, "minimum": 1.0, "maximum": 2.0, "step": 0.01}, + {"name": "min_p", "type": "slider", "label": "min_p", "label_en": "Min-p", "default": 0.05, "minimum": 0.0, "maximum": 1.0, "step": 0.01, "info_en": "Chatterbox's real truncation filter: tokens below max_prob * min_p are masked. top_p defaults to 1.0 and is a deliberate no-op."}, + {"name": "top_p", "type": "slider", "label": "top_p", "label_en": "Top-p", "default": 1.0, "minimum": 0.0, "maximum": 1.0, "step": 0.01, "info_en": "Inert at 1.0 by design; lower it only if you also raise min_p out of the way."}, + {"name": "s3gen_cfg_rate", "type": "slider", "label": "s3gen_cfg_rate", "label_en": "S3Gen guidance", "default": 0.7, "minimum": 0.0, "maximum": 2.0, "step": 0.05} ], "chatterbox-vc": [ @@ -69,13 +83,23 @@ {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 32, "minimum": 1, "step": 1, "precision": 0}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "speed", "type": "slider", "label": "speed", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05}, - {"name": "instruct", "type": "text", "label": "instruct（风格/音色指令）", "default": "", "placeholder": "如：以轻快的语气朗读"} + {"name": "instruct", "type": "text", "label": "instruct（风格/音色指令）", "default": "", "placeholder": "如：以轻快的语气朗读"}, + {"name": "t_shift", "type": "slider", "label": "t_shift", "label_en": "Timestep shift", "default": 0.1, "minimum": 0.0, "maximum": 1.0, "step": 0.01}, + {"name": "class_temperature", "type": "slider", "label": "class_temperature", "label_en": "Class temperature", "default": 0.0, "minimum": 0.0, "maximum": 2.0, "step": 0.05, "info_en": "0 picks the argmax token; above 0 samples the codebook class."}, + {"name": "position_temperature", "type": "slider", "label": "position_temperature", "label_en": "Position temperature", "default": 5.0, "minimum": 0.0, "maximum": 20.0, "step": 0.5, "info_en": "0 disables Gumbel sampling of the unmasking order."}, + {"name": "layer_penalty_factor", "type": "slider", "label": "layer_penalty_factor", "label_en": "Layer penalty", "default": 5.0, "minimum": 0.0, "maximum": 20.0, "step": 0.5, "info_en": "Score penalty per codebook layer when choosing which position to unmask next."}, + {"name": "denoise", "type": "bool", "label": "denoise", "label_en": "Denoise reference", "default": true}, + {"name": "preprocess_prompt", "type": "bool", "label": "preprocess_prompt", "label_en": "Preprocess reference", "default": true}, + {"name": "postprocess_output", "type": "bool", "label": "postprocess_output", "label_en": "Postprocess output", "default": true}, + {"name": "audio_chunk_threshold", "type": "number", "label": "audio_chunk_threshold", "label_en": "Long-form chunk threshold (s)", "default": 30.0, "minimum": 0.0, "step": 1.0}, + {"name": "audio_chunk_duration", "type": "number", "label": "audio_chunk_duration", "label_en": "Long-form chunk length (s)", "default": 15.0, "minimum": 0.0, "step": 1.0}, + {"name": "text_chunk_mode", "type": "choice", "label": "text_chunk_mode", "label_en": "Text chunk mode", "default": "tag_aware", "choices": ["default", "tag_aware", "japanese", "endline"]} ], "sense_asr": [ {"name": "enable_itn", "type": "bool", "label": "enable_itn（逆文本规范化）", "label_en": "enable_itn", "default": true}, {"name": "keep_tags", "type": "bool", "label": "keep_tags（保留语言/情绪/事件标签）", "label_en": "keep_tags", "default": false}, - {"name": "audio_chunk_mode", "type": "choice", "label": "audio_chunk_mode", "default": "auto", "choices": ["auto", "fixed", "none"]}, + {"name": "audio_chunk_mode", "type": "choice", "label": "audio_chunk_mode", "label_en": "Audio chunk mode", "default": "auto", "choices": ["auto", "fixed", "vad", "none"], "info_en": "vad segments on detected speech; auto and vad share the same code path."}, {"name": "audio_chunk_duration_sec", "type": "number", "label": "audio_chunk_duration_sec", "default": 30, "minimum": 0.001, "step": 1} ], @@ -85,7 +109,10 @@ "neutts": [ {"name": "voice_id", "type": "choice", "label": "voice_id（内置音色）", "label_en": "voice_id (built-in voice)", "default": "emily", "choices": ["dave", "emily", "greta", "jo", "juliette", "mateo", "paul", "sophie", "steven"]}, - {"name": "emotion", "type": "choice", "label": "emotion（情绪）", "label_en": "emotion", "default": "neutral", "choices": ["angry", "disgusted", "sad", "happy", "fearful", "neutral", "surprised"]} + {"name": "emotion", "type": "choice", "label": "emotion（情绪）", "label_en": "emotion", "default": "neutral", "choices": ["angry", "disgusted", "sad", "happy", "fearful", "neutral", "surprised"]}, + {"name": "temperature", "type": "slider", "label": "temperature", "label_en": "Temperature", "default": 1.0, "minimum": 0.05, "maximum": 2.0, "step": 0.05, "info_en": "Must be positive."}, + {"name": "top_k", "type": "number", "label": "top_k", "label_en": "Top-k", "default": 50, "minimum": 1, "step": 1, "precision": 0, "info_en": "Must be positive."}, + {"name": "min_tokens", "type": "number", "label": "min_tokens", "label_en": "Min speech tokens", "default": 50, "minimum": 0, "step": 1, "precision": 0, "info_en": "Minimum tokens generated before an end-of-speech token may stop decoding."} ], "magpie_tts": [ @@ -147,14 +174,20 @@ {"name": "flow_edit_n_avg", "type": "number", "label": "【remix】flow_edit_n_avg", "default": 2, "minimum": 1, "maximum": 4, "step": 1, "precision": 0, "info": "每步多次采样取平均（remix 默认 2）；1=最快"}, {"name": "bpm", "type": "number", "label": "【曲谱】BPM", "default": 0, "minimum": 0, "step": 1, "precision": 0, "info": "0=不指定"}, {"name": "keyscale", "type": "text", "label": "【曲谱】keyscale", "default": "", "placeholder": "如 F major"}, - {"name": "timesignature", "type": "text", "label": "【曲谱】timesignature", "default": "", "placeholder": "如 4"} + {"name": "timesignature", "type": "text", "label": "【曲谱】timesignature", "default": "", "placeholder": "如 4"}, + {"name": "negative_prompt", "type": "text", "label": "negative_prompt", "label_en": "Negative prompt", "default": "", "placeholder_en": "Blank uses the model's \"NO USER INPUT\" placeholder"} ], "minimax_music3": [ {"name": "num_inference_steps", "type": "number", "label": "Flow steps per window", "default": 30, "minimum": 1, "maximum": 200, "step": 1, "precision": 0, "info": "Flow-matching Euler steps per 200-frame denoising window."}, {"name": "guidance_scale", "type": "slider", "label": "Flow guidance scale", "default": 1.7, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, {"name": "ar_guidance_scale", "type": "slider", "label": "AR guidance scale", "default": 1.5, "minimum": 0.0, "maximum": 10.0, "step": 0.1, "info": "Classifier-free guidance of the semantic and residual code sampling."}, - {"name": "top_k", "type": "number", "label": "top_k", "default": 50, "minimum": 1, "maximum": 1024, "step": 1, "precision": 0} + {"name": "top_k", "type": "number", "label": "top_k", "default": 50, "minimum": 1, "maximum": 1024, "step": 1, "precision": 0}, + {"name": "ensemble_takes", "type": "number", "label": "ensemble_takes", "label_en": "Ensemble takes", "default": 1, "minimum": 1, "maximum": 16, "step": 1, "precision": 0, "info_en": "Decode N independent takes in one batched pass (seeds seed, seed+1, ...). Extra takes are returned as take_01...take_NN."}, + {"name": "ensemble_prefix_frames", "type": "number", "label": "ensemble_prefix_frames", "label_en": "Ensemble intro lock (frames)", "default": 0, "minimum": 0, "step": 10, "precision": 0, "info_en": "Decode the first N frames once and share them across takes before they diverge. 0 disables the fork."}, + {"name": "flow_uncond_interval", "type": "number", "label": "flow_uncond_interval", "label_en": "Flow CFG reuse interval", "default": 1, "minimum": 1, "step": 1, "precision": 0, "info_en": "Above 1 reuses the cached guidance delta on intermediate steps: faster, slightly off the exact reference trajectory. 1 is exact."}, + {"name": "flow_uncond_warmup", "type": "number", "label": "flow_uncond_warmup", "label_en": "Flow CFG warmup steps", "default": 2, "minimum": 0, "step": 1, "precision": 0, "info_en": "Leading steps that always evaluate the unconditional branch when reuse is on."}, + {"name": "flow_chunk_hop_frames", "type": "number", "label": "flow_chunk_hop_frames", "label_en": "Flow chunk hop (frames)", "default": 0, "minimum": 0, "step": 10, "precision": 0, "info_en": "0 uses the model config's hop of 100. A larger hop means fewer chunks and less double-denoising."} ], "minimax_h3": [ @@ -163,14 +196,17 @@ {"name": "guidance_scale", "type": "slider", "label": "Guidance scale", "default": 1.0, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "sampler", "type": "choice", "label": "Sampler", "default": "euler", "choices": ["euler", "res_multistep", "dpmpp_2m", "unipc"]}, {"name": "dit_acceleration", "type": "choice", "label": "DiT acceleration", "default": "none", "choices": ["none", "spectrum", "first_block_cache"], "info": "None uses the quality-first full-DiT path. Acceleration modes are experimental and may distort some outputs."}, - {"name": "return_video", "type": "bool", "label": "Decode video", "default": false, "info": "Disabled by default to reduce memory use and return audio only."} + {"name": "return_video", "type": "bool", "label": "Decode video", "default": false, "info": "Disabled by default to reduce memory use and return audio only."}, + {"name": "negative_prompt", "type": "text", "label": "negative_prompt", "label_en": "Negative prompt", "default": "", "placeholder_en": "Blank uses the model default"} ], "stable_audio": [ {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 8, "minimum": 1, "step": 1, "precision": 0, "info": "RF 扩散步数"}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 1.0, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "audio_input_kind", "type": "choice", "label": "audio_input_kind（仅上传源音频时生效）", "default": "init_audio", "choices": ["init_audio", "inpaint_audio"]}, - {"name": "init_noise_level", "type": "slider", "label": "init_noise_level（init_audio 强度）", "default": 1.0, "minimum": 0.0, "maximum": 1.0, "step": 0.05} + {"name": "init_noise_level", "type": "slider", "label": "init_noise_level（init_audio 强度）", "default": 1.0, "minimum": 0.0, "maximum": 1.0, "step": 0.05}, + {"name": "negative_prompt", "type": "text", "label": "negative_prompt", "label_en": "Negative prompt", "default": "", "placeholder_en": "Blank applies no negative conditioning"}, + {"name": "sampler", "type": "choice", "label": "sampler", "label_en": "Sampler", "default": "pingpong", "choices": ["pingpong", "euler"], "info_en": "Only pingpong and euler are accepted; the dpmpp entries in the CLI help text are rejected by the parser."} ], "seed_vc": [ @@ -179,7 +215,10 @@ {"name": "length_adjust", "type": "slider", "label": "length_adjust（时长伸缩）", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05}, {"name": "intelligibility_cfg_rate", "type": "slider", "label": "intelligibility_cfg_rate（仅 v2_vc）", "default": 0.7, "minimum": 0.0, "maximum": 1.0, "step": 0.05}, {"name": "similarity_cfg_rate", "type": "slider", "label": "similarity_cfg_rate（仅 v2_vc）", "default": 0.7, "minimum": 0.0, "maximum": 1.0, "step": 0.05}, - {"name": "inference_cfg_rate", "type": "slider", "label": "inference_cfg_rate（仅 v1 路径）", "default": 0.7, "minimum": 0.0, "maximum": 1.0, "step": 0.05} + {"name": "inference_cfg_rate", "type": "slider", "label": "inference_cfg_rate（仅 v1 路径）", "default": 0.7, "minimum": 0.0, "maximum": 1.0, "step": 0.05}, + {"name": "f0_condition", "type": "bool", "label": "f0_condition（仅 v1 路径）", "label_en": "f0_condition (v1 routes only)", "default": false, "info_en": "Enable F0 conditioning. Required for singing voice conversion on the v1_svc route."}, + {"name": "auto_f0_adjust", "type": "bool", "label": "auto_f0_adjust（仅 v1 路径）", "label_en": "auto_f0_adjust (v1 routes only)", "default": false, "info_en": "Shift the source pitch toward the target speaker's pitch level."}, + {"name": "semitone_shift", "type": "number", "label": "semitone_shift（仅 v1 路径）", "label_en": "semitone_shift (v1 routes only)", "default": 0, "minimum": -24, "maximum": 24, "step": 1, "precision": 0} ], "rvc": [ @@ -229,7 +268,9 @@ {"name": "guidance_scale", "type": "slider", "label": "guidance_scale（MuLa CFG）", "default": 1.5, "minimum": 0.0, "maximum": 5.0, "step": 0.1}, {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps（codec 步数）", "default": 10, "minimum": 1, "step": 1, "precision": 0}, {"name": "infinite_mode", "type": "bool", "label": "infinite_mode（长输出分段生成）", "default": false}, - {"name": "codec_guidance_scale", "type": "slider", "label": "codec_guidance_scale", "default": 1.25, "minimum": 0.0, "maximum": 5.0, "step": 0.05} + {"name": "codec_guidance_scale", "type": "slider", "label": "codec_guidance_scale", "default": 1.25, "minimum": 0.0, "maximum": 5.0, "step": 0.05}, + {"name": "text_chunk_size", "type": "number", "label": "text_chunk_size", "label_en": "Text chunk size (infinite mode)", "default": 4096, "minimum": 1, "step": 64, "precision": 0}, + {"name": "infinite_chunk_audio_duration_ms", "type": "number", "label": "infinite_chunk_audio_duration_ms", "label_en": "Infinite-mode chunk audio (ms)", "default": 240000, "minimum": 1, "step": 10000, "precision": 0, "info_en": "Maximum audio generated per infinite-mode chunk. Must be positive."} ], "index_tts2": [ @@ -239,13 +280,32 @@ {"name": "use_emotion_text", "type": "bool", "label": "use_emotion_text（从朗读文本推断情感）", "label_en": "use_emotion_text (infer from text)", "default": false}, {"name": "use_random_emotion", "type": "bool", "label": "use_random_emotion（随机情感）", "label_en": "use_random_emotion", "default": false}, {"name": "interval_silence_ms", "type": "number", "label": "interval_silence_ms（分段间静音）", "label_en": "interval_silence_ms", "default": 200, "minimum": 0, "step": 50, "precision": 0}, - {"name": "duration_factor", "type": "slider", "label": "duration_factor（语速/时长倍率，>1 更慢，<1 更快）", "label_en": "duration_factor (duration multiplier; >1 slower, <1 faster)", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05, "info": "对齐官方 IndexTTS2.5 的 duration_factor：缩放输出时长，不改变音色/内容", "info_en": "Matches official IndexTTS2.5 duration_factor: scales output duration without changing timbre or content"} + {"name": "duration_factor", "type": "slider", "label": "duration_factor（语速/时长倍率，>1 更慢，<1 更快）", "label_en": "duration_factor (duration multiplier; >1 slower, <1 faster)", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05, "info": "对齐官方 IndexTTS2.5 的 duration_factor：缩放输出时长，不改变音色/内容", "info_en": "Matches official IndexTTS2.5 duration_factor: scales output duration without changing timbre or content"}, + {"name": "do_sample", "type": "bool", "label": "do_sample", "label_en": "Sample", "default": true}, + {"name": "temperature", "type": "slider", "label": "temperature", "label_en": "Temperature", "default": 0.8, "minimum": 0.05, "maximum": 2.0, "step": 0.05, "info_en": "Must be positive."}, + {"name": "top_p", "type": "slider", "label": "top_p", "label_en": "Top-p", "default": 0.8, "minimum": 0.01, "maximum": 1.0, "step": 0.01, "info_en": "Must be within (0, 1]."}, + {"name": "top_k", "type": "number", "label": "top_k", "label_en": "Top-k", "default": 30, "minimum": 1, "step": 1, "precision": 0, "info_en": "Must be positive."}, + {"name": "repetition_penalty", "type": "slider", "label": "repetition_penalty", "label_en": "Repetition penalty", "default": 10.0, "minimum": 0.0, "maximum": 20.0, "step": 0.1}, + {"name": "num_beams", "type": "number", "label": "num_beams", "label_en": "Beam count", "default": 3, "minimum": 1, "step": 1, "precision": 0, "info_en": "Must be positive."}, + {"name": "length_penalty", "type": "slider", "label": "length_penalty", "label_en": "Length penalty", "default": 0.0, "minimum": -2.0, "maximum": 2.0, "step": 0.1} ], "irodori_tts": [ {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps（RF 扩散步数）", "label_en": "num_inference_steps", "default": 40, "minimum": 1, "step": 1, "precision": 0}, {"name": "duration_sec", "type": "number", "label": "duration_sec（0=模型自动预测时长）", "label_en": "duration_sec (0 = auto)", "default": 0, "minimum": 0, "step": 0.5}, - {"name": "duration_scale", "type": "slider", "label": "duration_scale（语速倒数，越大越慢）", "label_en": "duration_scale", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05} + {"name": "duration_scale", "type": "slider", "label": "duration_scale（语速倒数，越大越慢）", "label_en": "duration_scale", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "step": 0.05}, + {"name": "text_guidance_scale", "type": "slider", "label": "text_guidance_scale", "label_en": "Text guidance", "default": 3.0, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, + {"name": "speaker_guidance_scale", "type": "slider", "label": "speaker_guidance_scale", "label_en": "Speaker guidance", "default": 5.0, "minimum": 0.0, "maximum": 15.0, "step": 0.1}, + {"name": "caption_guidance_scale", "type": "slider", "label": "caption_guidance_scale", "label_en": "Caption guidance", "default": 3.0, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, + {"name": "guidance_mode", "type": "choice", "label": "guidance_mode", "label_en": "Guidance mode", "default": "independent", "choices": ["independent", "joint", "alternating"]}, + {"name": "guidance_min_t", "type": "slider", "label": "guidance_min_t", "label_en": "Guidance start timestep", "default": 0.5, "minimum": 0.0, "maximum": 1.0, "step": 0.05, "info_en": "Must not exceed the guidance end timestep."}, + {"name": "guidance_max_t", "type": "slider", "label": "guidance_max_t", "label_en": "Guidance end timestep", "default": 1.0, "minimum": 0.0, "maximum": 1.0, "step": 0.05}, + {"name": "min_duration_sec", "type": "number", "label": "min_duration_sec", "label_en": "Min duration (s)", "default": 0.5, "minimum": 0.05, "step": 0.5, "info_en": "Must be positive and no greater than the maximum."}, + {"name": "max_duration_sec", "type": "number", "label": "max_duration_sec", "label_en": "Max duration (s)", "default": 30.0, "minimum": 0.05, "step": 0.5}, + {"name": "no_ref", "type": "bool", "label": "no_ref", "label_en": "No-reference generation", "default": true, "info_en": "Forced off automatically whenever a speaker reference is supplied."}, + {"name": "trim_tail", "type": "bool", "label": "trim_tail", "label_en": "Trim trailing silence", "default": true}, + {"name": "text_chunk_mode", "type": "choice", "label": "text_chunk_mode", "label_en": "Text chunk mode", "default": "endline", "choices": ["japanese", "endline"]}, + {"name": "instruction", "type": "text", "label": "instruction", "label_en": "instruction (caption-conditioned checkpoints)", "default": "", "placeholder_en": "Voice or style description; ignored by checkpoints without caption conditioning"} ], "moss_tts_local": [ @@ -320,7 +380,8 @@ {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0}, {"name": "top_k", "type": "number", "label": "top_k", "default": 20, "minimum": 0, "step": 1, "precision": 0}, {"name": "top_p", "type": "slider", "label": "top_p", "default": 0.8, "minimum": 0.0, "maximum": 1.0, "step": 0.01}, - {"name": "temperature", "type": "slider", "label": "temperature", "default": 0.7, "minimum": 0.0, "maximum": 2.0, "step": 0.05} + {"name": "temperature", "type": "slider", "label": "temperature", "default": 0.7, "minimum": 0.0, "maximum": 2.0, "step": 0.05}, + {"name": "min_new_audio_steps", "type": "number", "label": "min_new_audio_steps", "label_en": "Min audio steps", "default": 6, "minimum": 0, "step": 1, "precision": 0, "info_en": "Floor on generated audio steps; suppresses truncated outputs. Must be non-negative."} ], "firered-audio-vdesign": [ @@ -328,7 +389,8 @@ {"name": "language", "type": "choice", "label": "language", "default": "zh", "choices": ["zh", "en"]}, {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 10, "minimum": 1, "step": 1, "precision": 0}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, - {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0} + {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0}, + {"name": "min_new_audio_steps", "type": "number", "label": "min_new_audio_steps", "label_en": "Min audio steps", "default": 6, "minimum": 0, "step": 1, "precision": 0, "info_en": "Floor on generated audio steps; suppresses truncated outputs. Must be non-negative."} ], "firered-audio-semantic-edit": [ @@ -338,7 +400,8 @@ {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 10, "minimum": 1, "step": 1, "precision": 0}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0}, - {"name": "max_new_text_tokens", "type": "number", "label": "max_new_text_tokens", "default": 512, "minimum": 1, "step": 1, "precision": 0} + {"name": "max_new_text_tokens", "type": "number", "label": "max_new_text_tokens", "default": 512, "minimum": 1, "step": 1, "precision": 0}, + {"name": "min_new_audio_steps", "type": "number", "label": "min_new_audio_steps", "label_en": "Min audio steps", "default": 6, "minimum": 0, "step": 1, "precision": 0, "info_en": "Floor on generated audio steps; suppresses truncated outputs. Must be non-negative."} ], "firered-audio-acoustic-edit": [ @@ -347,7 +410,8 @@ {"name": "instruction", "type": "text", "label": "instruction", "default": "shift the pitch by 3 steps", "placeholder": "shift the pitch by 3 steps"}, {"name": "num_inference_steps", "type": "number", "label": "num_inference_steps", "default": 10, "minimum": 1, "step": 1, "precision": 0}, {"name": "guidance_scale", "type": "slider", "label": "guidance_scale", "default": 2.0, "minimum": 0.0, "maximum": 10.0, "step": 0.1}, - {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0} + {"name": "max_new_audio_steps", "type": "number", "label": "max_new_audio_steps", "default": 750, "minimum": 1, "step": 1, "precision": 0}, + {"name": "min_new_audio_steps", "type": "number", "label": "min_new_audio_steps", "label_en": "Min audio steps", "default": 6, "minimum": 0, "step": 1, "precision": 0, "info_en": "Floor on generated audio steps; suppresses truncated outputs. Must be non-negative."} ], "firered-audio-asr": [