Functions — github.com/ASLP-lab/SenSE

CodeHub MCPcopy Index your code

hub / github.com/ASLP-lab/SenSE / functions

Functions360 in github.com/ASLP-lab/SenSE

⨍ Functions 360◇ Types & classes 78

(self, text_num_embeds, text_dim, mask_padding=True, conv_layers=0, conv_mult=2)

src/sense/model/backbones/unett.py:36

(self, mel_dim, text_dim, out_dim)

src/sense/model/backbones/unett.py:89

(self, text_num_embeds, text_dim, mask_padding=True, conv_layers=0, conv_mult=2)

src/sense/model/backbones/dit.py:32

(self, mel_dim, text_dim, out_dim)

src/sense/model/backbones/dit.py:85

(self, cfg=None)

src/sense/model/encoder/wavlm/wavlm.py:165

( self, conv_layers: List[Tuple[int, int, int]], dropout: float = 0.0,

src/sense/model/encoder/wavlm/wavlm.py:429

src/sense/model/encoder/wavlm/wavlm.py:558

( self, embedding_dim: float = 768, ffn_embedding_dim: float = 3072,

src/sense/model/encoder/wavlm/wavlm.py:672

(self, config, model)

src/sense/model/encoder/wavlm/wavlm_encoder.py:8

( self, input_dim: int = 80, encoder_dim: int = 512, num_layer

src/sense/model/encoder/conformer/encoder2.py:138

(self, module: nn.Module, module_factor: float = 1.0, input_factor: float = 1.0)

src/sense/model/encoder/conformer/modules.py:26

(self, in_features: int, out_features: int, bias: bool = True)

src/sense/model/encoder/conformer/modules.py:41

(self, shape: tuple)

src/sense/model/encoder/conformer/modules.py:68

(self, d_model: int, num_heads: int, dropout_p: float = 0.1)

src/sense/model/encoder/conformer/attention.py:136

( self, in_channels: int, out_channels: int, stride: int = 1,

src/sense/model/encoder/conformer/convolution.py:86

( self, in_channels: int, kernel_size: int = 31, expansion_fac

src/sense/model/encoder/conformer/convolution.py:125

(self, in_channels: int, out_channels: int)

src/sense/model/encoder/conformer/convolution.py:167

(self, in_channels: int, out_channels: int)

src/sense/model/encoder/conformer/convolution.py:203

( self, input_dim: int = 80, encoder_dim: int = 512, num_layer

src/sense/model/encoder/conformer/encoder.py:138

src/sense/model/encoder/conformer/activation.py:24

(self, d_model: int = 512, max_len: int = 5000)

src/sense/model/encoder/conformer/embedding.py:29

( self, encoder_dim: int = 512, expansion_factor: int = 4, dro

src/sense/model/encoder/conformer/feed_forward.py:40

(self, channels, kernel_size=1, stride=1, padding=0, dilation=1, bias=True, scale=4)

src/sense/eval/ecapa_tdnn.py:22

(self, in_channels, out_channels, kernel_size=1, stride=1, padding=0, dilation=1, bias=True)

src/sense/eval/ecapa_tdnn.py:61

(self, in_channels, out_channels, kernel_size, stride, padding, dilation, scale, se_bottleneck_dim)

src/sense/eval/ecapa_tdnn.py:102

(self, in_dim, attention_channels=128, global_context_att=False)

src/sense/eval/ecapa_tdnn.py:135

( self, feat_dim=80, channels=512, emb_dim=192, global_context_att=Fal

src/sense/eval/ecapa_tdnn.py:165

(self, primary_model_path, p808_model_path)

src/sense/eval/eval_dnsmos_oval.py:19

(self, primary_model_path, p808_model_path)

src/sense/eval/eval_dnsmos.py:19

src/sense/eval/intrusive_se_metrics.py:26

Args: sr (int): Sampling rate. model_type (str): Model type. Select from "hubert-base", "hubert-large", "wav2vec2-bas

src/sense/eval/SpeechBERTScore/speechbertscore.py:41

src/sense/model/dataset.py:347

src/sense/model/dataset.py:133

src/sense/model/dataset.py:224

src/sense/model/dataset.py:359

Function_forward_pre_hook

src/sense/model/modules2.py:249

Method_remove_weight_norm_from_wavlm

Remove weight_norm from WavLM to avoid deep copy issues

src/sense/model/llama_llm.py:244

Functionadd_reverberation_v2

(speech_sample, rir_sample, fs)

src/sense/model/dataset.py:77

Functionalign_audio_duration

(noise_audio, target_audio)

src/sense/model/dataset.py:20

Methodaudio_melspec

(self, audio, n_mels=120, frame_size=320, hop_length=160, sr=16000, to_db=True)

src/sense/eval/eval_dnsmos_oval.py:23

( n_in, n_out, k, stride, is_l

src/sense/model/encoder/wavlm/wavlm.py:441

Functioncalculate_dnsmos_score

(testset_dir, dnsmos_path, csv_path=None, json_path=None)

src/sense/eval/eval_dnsmos_oval.py:139

Functioncalculate_dnsmos_score

(testset_dir, dnsmos_path, csv_path=None, json_path=None)

src/sense/eval/eval_dnsmos.py:158

Functionchunk_text

Splits the input text into chunks, each with a maximum number of characters. Args: text (str): The text to be split. max_cha

src/sense/infer/utils_infer.py:68

Methodckpt_forward

src/sense/model/backbones/dit.py:177

Methodclear_cache

src/sense/model/backbones/mmdit.py:144

Methodclear_cache

src/sense/model/backbones/unett.py:181

Functioncollate_fn

src/sense/model/dataset.py:472

Functiondeal_with_audio_dir

src/sense/train/datasets/prepare_libritts_noisy.py:17

Functiondeal_with_audio_dir

src/sense/train/datasets/prepare_libritts.py:17

Functiondeal_with_audio_dir

(audio_dir, token_dict)

src/sense/train/datasets/prepare_libritts_noisy_token.py:54

Functiondeal_with_audio_dir

src/sense/train/datasets/prepare_emilia.py:111

Functiondeal_with_audio_dir

(audio_dir, audio_jsonl)

src/sense/train/datasets/prepare_emilia_se.py:146

Functiondeal_with_sub_path_files

(dataset_path, sub_path)

src/sense/train/datasets/prepare_wenetspeech4tts.py:21

src/sense/model/cfm.py:84

src/sense/model/cfm.py:177

src/sense/model/modules2.py:23

src/sense/model/modules2.py:33

src/sense/model/modules2.py:48

(ctx, x, scale)

src/sense/model/modules2.py:61

src/sense/model/modules2.py:79

src/sense/model/modules2.py:94

src/sense/model/modules2.py:119

Input shape: Time x Batch x Channel Args: key_padding_mask (ByteTensor, optional): mask to exclude keys that are

src/sense/model/modules2.py:455

( self, inp: float["b n d"] | float["b nw"], # mel or raw wave # noqa: F722 noisy_in

src/sense/model/cfm.py:231

src/sense/model/modules.py:261

(self, x, scale=1000)

src/sense/model/modules.py:285

(self, x: float["b n d"], mask: bool["b n"] | None = None)

src/sense/model/modules.py:309

src/sense/model/modules.py:362

(self, x: torch.Tensor)

src/sense/model/modules.py:390

src/sense/model/modules.py:413

(self, x, emb=None)

src/sense/model/modules.py:441

src/sense/model/modules.py:462

src/sense/model/modules.py:483

( self, x: float["b n d"], # noised input x # noqa: F722 c: float["b n d"] = None,

src/sense/model/modules.py:549

(self, x, t, mask=None, rope=None)

src/sense/model/modules.py:758

(self, x, c, t, mask=None, rope=None, c_rope=None)

src/sense/model/modules.py:818

(self, timestep: float["b"])

src/sense/model/modules.py:858

src/sense/model/projector.py:14

src/sense/model/projector.py:40

(self, x, atts)

src/sense/model/projector.py:69

(self, mels, mels_len, speech_tokens, speech_tokens_lengths, wavs=None, wavs_len=None)

src/sense/model/llama_llm.py:99

(self, text: int["b nt"], drop_text=False)

src/sense/model/backbones/mmdit.py:39

(self, x: float["b n d"], cond: float["b n d"], drop_audio_cond=False)

src/sense/model/backbones/mmdit.py:72

( self, x: float["b n d"], # nosied input audio # noqa: F722 cond: float["b n d"],

src/sense/model/backbones/mmdit.py:147

(self, text: int["b nt"], seq_len, drop_text=False)

src/sense/model/backbones/unett.py:52

(self, x: float["b n d"], cond: float["b n d"], text_embed: float["b n d"], drop_audio_cond=False)

src/sense/model/backbones/unett.py:94

( self, x: float["b n d"], # nosied input audio # noqa: F722 cond: float["b n d"],

src/sense/model/backbones/unett.py:184

(self, text: int["b nt"], seq_len, drop_text=False)

src/sense/model/backbones/dit.py:48

(self, x: float["b n d"], cond: float["b n d"], cond_noisy: float["b n d"], text_embed: float["b n d"], drop_a

src/sense/model/backbones/dit.py:90

( self, x: float["b n d"], # nosied input audio # noqa: F722 cond: float["b n d"],

src/sense/model/backbones/dit.py:186

(self, x, mask=None)

src/sense/model/encoder/wavlm/wavlm.py:535

(self, x, padding_mask=None, streaming_mask=None, layer=None)

src/sense/model/encoder/wavlm/wavlm.py:614

LayerNorm is applied either before or after the self-attention/ffn modules similar to the original Transformer imlementation.

src/sense/model/encoder/wavlm/wavlm.py:728

(self, inputs: Tensor)

src/sense/model/encoder/conformer/encoder2.py:108

Forward propagate a `inputs` for encoder training. Args: inputs (torch.FloatTensor): A input sequence passed to encoder

src/sense/model/encoder/conformer/encoder2.py:182

(self, inputs: Tensor)

src/sense/model/encoder/conformer/modules.py:32

(self, x: Tensor)

src/sense/model/encoder/conformer/modules.py:48

(self, x: Tensor)

src/sense/model/encoder/conformer/modules.py:59

(self, x: Tensor)

src/sense/model/encoder/conformer/modules.py:72

← previous next →201–300 of 360, ranked by callers