Functions — github.com/OpenMOSS/MOSS-TTS

hub / github.com/OpenMOSS/MOSS-TTS / functions

Functions1,298 in github.com/OpenMOSS/MOSS-TTS

⨍ Functions 1,298◇ Types & classes 178↳ Endpoints 16

↓ 2 callersFunction_resample_wav

(wav: torch.Tensor, orig_sr: int, target_sr: int)

clis/moss_ttsd_app.py:150

↓ 2 callersFunction_resolve_audio_path

(audio_path: str | None, use_default: bool, default_path: str | Path)

moss_tts_realtime/app.py:440

↓ 2 callersFunction_resolve_audio_path

( audio_path: Optional[str], use_default: bool, default_path: str, )

moss_tts_realtime/fast_api.py:307

↓ 2 callersMethod_resolve_fixed_nq

( self, n_vq_for_inference: Optional[int] = None, nq: Optional[int] = None, )

moss_tts_local_v1.5/modeling_moss_tts.py:163

↓ 2 callersFunction_resolve_ggml_type

Map a human-readable type name (e.g. 'q8_0') to its ggml_type int.

moss_tts_delay/llama_cpp/backbone.py:92

↓ 2 callersMethod_sample_next_assistant_text_token

( self, local_hidden_states: torch.FloatTensor, do_sample: bool, temperature:

moss_tts_local_v1.5/modeling_moss_tts.py:415

↓ 2 callersMethod_shard_paths

(self, shard_id: int)

moss_soundeffect_v2/diffsynth/trainers/cache_shards.py:194

↓ 2 callersFunction_status_from_snapshot

(snapshot: WarmupSnapshot)

moss_tts_realtime/app.py:858

↓ 2 callersMethod_stream_chunk

(self, frames: list[np.ndarray], callback)

moss_tts_delay/llama_cpp/pipeline.py:753

↓ 2 callersMethod_tokenize

(self, text: str)

moss_tts_realtime/mossttsrealtime/streaming_mossttsrealtime.py:649

↓ 2 callersMethod_validate_code_list

( self, codes_list: Optional[List[Optional[torch.Tensor]]], target_n_vq: int,

moss_tts_local_v1.5/finetuning/dataset.py:98

↓ 2 callersFunction_validate_dialogue_text

(dialogue_text: str, speaker_count: int)

clis/moss_ttsd_app.py:237

↓ 2 callersFunction_validate_source_dir

(arg_name: str, model_dir: Path)

scripts/fuse_moss_tts_delay_with_codec.py:755

↓ 2 callersFunction_warmup_gate_message

(snapshot: WarmupSnapshot)

moss_tts_realtime/app.py:851

↓ 2 callersFunctionapply_rotary_pos_emb

( hidden_states: torch.Tensor, cos: torch.Tensor, sin: torch.Tensor, )

moss_tts_local_v1.5/gpt2_decoder.py:76

↓ 2 callersFunctionapply_top_p

Nucleus sampling: keep tokens whose cumulative probability <= top_p.

moss_tts_delay/llama_cpp/sampling.py:24

↓ 2 callersFunctionbatch_encode

( processor: MossTTSDelayProcessor, paths: List[str], batch_size: int, n_vq: Optional[int],

moss_tts_local/finetuning/prepare_data.py:99

↓ 2 callersFunctionbatch_encode

( processor: MossTTSDelayProcessor, paths: List[str], batch_size: int, n_vq: Optional[int],

moss_tts_delay/finetuning/prepare_data.py:99

↓ 2 callersFunctionbatch_encode

( processor, paths: List[str], batch_size: int, n_vq: Optional[int], desc: str, )

moss_tts_local_v1.5/finetuning/prepare_data.py:188

↓ 2 callersMethodbuild_assistant_message

( audio_codes_list: List[Union[str, torch.Tensor]], content: str = AUDIO_PLACEHOLDER, )

moss_tts_delay/processing_moss_tts.py:401

↓ 2 callersMethodbuild_packed_metadata

( hidden_states: torch.Tensor, cu_seqlens: torch.Tensor, num_sequences: Optional[torch

moss_tts_local_v1.5/gpt2_decoder.py:566

↓ 2 callersMethodbuild_packed_position_ids

( attention_mask: Optional[torch.Tensor], cu_seqlens: torch.Tensor, num_sequences: Opt

moss_tts_local_v1.5/gpt2_decoder.py:541

↓ 2 callersFunctionbuild_processor_config

(fused_config: JSONDict, tokenizer_json_path: Path)

scripts/fuse_moss_tts_delay_with_codec.py:551

↓ 2 callersFunctionbuild_processor_inputs

( runtime: StreamingRuntime, request: StreamingRequest, )

moss_tts_local_v1.5/streaming.py:327

↓ 2 callersFunctionbuild_remapped_main_weight_map

(main_weight_map: dict[str, str])

scripts/fuse_moss_tts_delay_with_codec.py:247

↓ 2 callersMethodbuild_user_message

( text: Optional[str] = None, reference: Optional[List[Optional[Union[str, torch.Tensor]]]] =

moss_tts_delay/processing_moss_tts.py:375

↓ 2 callersFunctioncollect_tokenizer_asset_filenames

(model_dir: Path)

scripts/fuse_moss_tts_delay_with_codec.py:493

↓ 2 callersMethodcrop_and_resize

(self, image, target_height, target_width)

moss_soundeffect_v2/diffsynth/trainers/utils.py:234

↓ 2 callersMethoddecode_code

(self, embed_id)

moss_soundeffect_v2/diffsynth/models/dac_vae.py:402

↓ 2 callersMethoddecode_latents

(self, latents)

moss_soundeffect_v2/diffsynth/models/dac_vae.py:405

↓ 2 callersMethodencode

(self, text: str)

moss_tts_delay/llama_cpp/processor.py:49

↓ 2 callersMethodencode_prompt

(self, prompt, positive=True, device="cuda")

moss_soundeffect_v2/diffsynth/prompters/wan_prompter.py:98

↓ 2 callersFunctionexport_finetuned_to_hf

Rebuild a public HF dir from a fine-tuned DiT checkpoint. Only the DiT is replaced — VAE / text_encoder / tokenizer / scheduler / package sou

moss_soundeffect_v2/hf_export.py:98

↓ 2 callersMethodexport_trainable_state_dict

(self, state_dict, remove_prefix=None)

moss_soundeffect_v2/diffsynth/trainers/utils.py:762

↓ 2 callersFunctionextract_audio_segments

Extract non-padding audio segments after de-delaying.

moss_tts_delay/llama_cpp/delay_state.py:286

↓ 2 callersFunctionfind_last_equal_C

tensor: torch.Tensor of shape [batch_size, seq_len] C: scalar value to match Returns: torch.Tensor of shape [batch_size] with last indice

moss_tts_delay/inference_utils.py:148

↓ 2 callersMethodfinish

( self, max_steps: Optional[int] = None, temperature: float = 0.8, top_p: floa

moss_tts_realtime/mossttsrealtime/streaming_mossttsrealtime.py:304

↓ 2 callersFunctionflash_attention

(q: torch.Tensor, k: torch.Tensor, v: torch.Tensor, num_heads: int, compatibility_mode=False)

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:36

↓ 2 callersFunctionflush_backbone_shard

()

moss_tts_delay/llama_cpp/conversion/extract_weights.py:124

↓ 2 callersFunctionflush_backbone_shard

()

scripts/extract_weights_llama_cpp.py:112

↓ 2 callersFunctionflush_shard

()

scripts/fuse_moss_tts_delay_with_codec.py:303

↓ 2 callersMethodforward_preprocess

(self, data)

moss_soundeffect_v2/finetuning/finetuning.py:90

↓ 2 callersMethodfrom_yaml

(cls, path: str | Path)

moss_tts_delay/llama_cpp/pipeline.py:258

↓ 2 callersMethodgenerate_local_transformer

( self, hidden_states: torch.Tensor, temperature: float, top_p: float,

moss_tts_realtime/inferencer.py:354

↓ 2 callersMethodgenerate_local_transformer

( self, hidden_states: torch.Tensor, temperature: float, top_p: float,

moss_tts_realtime/mossttsrealtime/streaming_mossttsrealtime.py:331

↓ 2 callersMethodgenerate_noise

(self, shape, seed=None, rand_device="cpu", rand_torch_dtype=torch.float32, device=None, torch_dtype=None)

moss_soundeffect_v2/diffsynth/utils/__init__.py:116

↓ 2 callersMethodget_height_width

(self, image)

moss_soundeffect_v2/diffsynth/trainers/utils.py:246

↓ 2 callersMethodget_or_load_backend

(self, backend: BackendPaths)

moss_tts_realtime/app.py:453

↓ 2 callersMethodget_or_load_backend

(self, backend: BackendPaths)

moss_tts_realtime/fast_api.py:331

↓ 2 callersMethodget_output_length

(self, input_length)

moss_soundeffect_v2/diffsynth/models/dac_vae.py:111

↓ 2 callersFunctiongradient_checkpointing_modules

(scope: str)

moss_tts_local_v1.5/finetuning/sft.py:518

↓ 2 callersFunctionis_finite_scalar

(value: torch.Tensor)

moss_tts_local_v1.5/finetuning/sft.py:858

↓ 2 callersFunctioniter_generate_frames

( model: Any, *, input_ids: torch.LongTensor, attention_mask: Optional[torch.Tensor], max_

moss_tts_local_v1.5/streaming.py:387

↓ 2 callersFunctionload_backend

(model_path: str, device_str: str, attn_implementation: str)

clis/moss_tts_app.py:117

↓ 2 callersFunctionload_backend

(model_path: str, device_str: str, attn_implementation: str)

clis/moss_voice_generator_app.py:104

↓ 2 callersFunctionload_backend

(model_path: str, codec_path: str, device_str: str, attn_implementation: str)

clis/moss_ttsd_app.py:118

↓ 2 callersFunctionload_backend

(model_dir: str, device_str: str)

clis/moss_sound_effect_app.py:27

↓ 2 callersMethodload_data

(self, file_path)

moss_soundeffect_v2/diffsynth/trainers/utils.py:127

↓ 2 callersMethodload_data

(self, file_path, start_time=None, end_time=None)

moss_soundeffect_v2/diffsynth/trainers/utils.py:575

↓ 2 callersFunctionload_runtime

( *, model_dir: str | Path = DEFAULT_MODEL_DIR, codec_dir: str | Path = DEFAULT_CODEC_DIR, dev

moss_tts_local_v1.5/streaming.py:175

↓ 2 callersFunctionload_source_index

(model_dir: Path)

scripts/fuse_moss_tts_delay_with_codec.py:195

↓ 2 callersFunctionlocal_processor_source_path

()

scripts/fuse_moss_tts_delay_with_codec.py:536

↓ 2 callersFunctionmain

()

moss_tts_delay/llama_cpp/pipeline.py:865

↓ 2 callersMethodmake_ensemble

(self, prompt_audio_tokens: Optional[np.ndarray] = None)

moss_tts_realtime/mossttsrealtime/processing_mossttsrealtime.py:107

↓ 2 callersMethodmode

(self)

moss_soundeffect_v2/diffsynth/models/dac_vae.py:600

↓ 2 callersFunctionmodel_load_context

()

moss_tts_local_v1.5/finetuning/sft.py:448

↓ 2 callersFunctionmodulate

(x: torch.Tensor, shift: torch.Tensor, scale: torch.Tensor)

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:76

↓ 2 callersFunctionmultinomial

Sample one token per row via multinomial distribution.

moss_tts_delay/llama_cpp/sampling.py:94

↓ 2 callersFunctionnormalize_audio_code_list

( value: Any, field_name: str, *, allow_none: bool = False, )

moss_tts_local_v1.5/finetuning/dataset.py:30

↓ 2 callersFunctionnormalize_language_tag

(language_tag: str | None)

clis/moss_tts_app.py:241

↓ 2 callersFunctionnormalize_text

(text: str)

clis/moss_ttsd_app.py:179

↓ 2 callersMethodpatchify

(self, x: torch.Tensor, control_camera_latents_input: Optional[torch.Tensor] = None)

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:370

↓ 2 callersFunctionprepare_batch

( batch: Dict[str, Any], device: torch.device, default_channelwise_weight: torch.Tensor, )

community/norwegian-lora/train_lora.py:308

↓ 2 callersMethodpreprocess

(self, audio_data, sample_rate)

moss_soundeffect_v2/diffsynth/models/dac_vae.py:877

↓ 2 callersFunctionpush_text

(session_id: str, text: str, is_final: bool)

moss_tts_realtime/tts_client.py:122

↓ 2 callersFunctionrender_mode_hint

(reference_audio: str | None, mode_with_reference: str)

clis/moss_tts_app.py:297

↓ 2 callersFunctionrepeat_kv

(hidden_states: torch.Tensor, n_rep: int)

moss_tts_realtime/mossttsrealtime/modeling_mossttsrealtime_local.py:87

↓ 2 callersFunctionresolve_attn_implementation

(requested: str, device: torch.device, dtype: torch.dtype)

clis/moss_tts_app.py:147

↓ 2 callersFunctionresolve_attn_implementation

(requested: str, device: torch.device, dtype: torch.dtype)

clis/moss_voice_generator_app.py:76

↓ 2 callersFunctionresolve_attn_implementation

(requested: str, device: torch.device, dtype: torch.dtype)

clis/moss_ttsd_app.py:90

↓ 2 callersFunctionresolve_attn_implementation

(requested: str, device: torch.device, dtype: torch.dtype)

moss_tts_local_v1.5/finetuning/sft.py:275

↓ 2 callersFunctionresolve_jsonl_paths

(spec: str | Sequence[str])

moss_tts_realtime/finetuning/common.py:34

↓ 2 callersFunctionresolve_jsonl_paths

(spec: str | Sequence[str])

moss_tts_local/finetuning/common.py:34

↓ 2 callersFunctionresolve_jsonl_paths

(spec: str | Sequence[str])

moss_tts_delay/finetuning/common.py:34

↓ 2 callersFunctionresolve_jsonl_paths

(spec: str | Sequence[str])

moss_tts_local_v1.5/finetuning/common.py:27

↓ 2 callersFunctionrope_apply

(x, freqs, num_heads)

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:104

↓ 2 callersFunctionrotate_half

(x)

moss_tts_realtime/mossttsrealtime/modeling_mossttsrealtime_local.py:73

↓ 2 callersFunctionrotate_half

(hidden_states: torch.Tensor)

moss_tts_local_v1.5/qwen3_decoder.py:80

↓ 2 callersFunctionsave_adapter_checkpoint

( model: torch.nn.Module, output_dir: str, tag: str, accelerator: Accelerator, )

community/norwegian-lora/train_lora.py:384

↓ 2 callersFunctionsave_checkpoint

( *, accelerator: Accelerator, model, model_path: str, codec_path: str, output_dir: Pa

moss_tts_local_v1.5/finetuning/sft.py:770

↓ 2 callersFunctionsearch_parameter

(param, state_dict)

moss_soundeffect_v2/diffsynth/models/utils.py:101

↓ 2 callersFunctionset_rank_cuda_device

(accelerator: Accelerator)

moss_tts_local_v1.5/finetuning/sft.py:241

↓ 2 callersFunctionsoftmax

Numerically stable softmax along last axis.

moss_tts_delay/llama_cpp/sampling.py:87

↓ 2 callersMethodstate_dict_converter

()

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:443

↓ 2 callersFunctionsupports_duration_control

(mode_with_reference: str)

clis/moss_tts_app.py:182

↓ 2 callersFunctionsynthesize_stream

( runtime: StreamingRuntime, request: StreamingRequest, *, output_dir: str | Path = DEFAULT_OU

moss_tts_local_v1.5/streaming.py:735

↓ 2 callersMethodtouch

(self)

moss_tts_realtime/fast_api.py:478

↓ 2 callersFunctiontransformers_zero3_load_context

(accelerator: Accelerator)

moss_tts_local_v1.5/finetuning/sft.py:364

↓ 2 callersMethodunpatchify

(self, x: torch.Tensor, grid_size: torch.Tensor)

moss_soundeffect_v2/diffsynth/models/wan_video_dit.py:380

↓ 2 callersFunctionunwrap_training_model

(model)

moss_tts_local_v1.5/finetuning/sft.py:511

← previous next →201–300 of 1,298, ranked by callers