Spaces:
Running
Running
File size: 3,520 Bytes
1d8403e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 | # ==================================================================================================
# DEEPFAKE AUDIO - vocoder/inference.py (Neural Waveform Synthesizer)
# ==================================================================================================
#
# 📝 DESCRIPTION
# This module provides the high-level API for vocoder inference. It encapsulates
# the WaveRNN model, handles hardware acceleration (CUDA), and provides the
# entry point for transforming Mel-Spectrograms into audible speech waveforms.
#
# 👤 AUTHORS
# - Amey Thakur (https://github.com/Amey-Thakur)
# - Mega Satish (https://github.com/msatmod)
#
# 🤝🏻 CREDITS
# Original Real-Time Voice Cloning methodology by CorentinJ
# Repository: https://github.com/CorentinJ/Real-Time-Voice-Cloning
#
# 🔗 PROJECT LINKS
# Repository: https://github.com/Amey-Thakur/DEEPFAKE-AUDIO
# Video Demo: https://youtu.be/i3wnBcbHDbs
# Research: https://github.com/Amey-Thakur/DEEPFAKE-AUDIO/blob/main/DEEPFAKE-AUDIO.ipynb
#
# 📜 LICENSE
# Released under the MIT License
# Release Date: 2021-02-06
# ==================================================================================================
from vocoder.models.fatchord_version import WaveRNN
from vocoder import hparams as hp
import torch
_model = None # Global singleton for the active WaveRNN instance
def load_model(weights_fpath, verbose=True):
"""Neural Wake-up: Initializes the WaveRNN architecture and loads pre-trained weights."""
global _model, _device
if verbose:
print("Building Wave-RNN Architecture...")
_model = WaveRNN(
rnn_dims=hp.voc_rnn_dims,
fc_dims=hp.voc_fc_dims,
bits=hp.bits,
pad=hp.voc_pad,
upsample_factors=hp.voc_upsample_factors,
feat_dims=hp.num_mels,
compute_dims=hp.voc_compute_dims,
res_out_dims=hp.voc_res_out_dims,
res_blocks=hp.voc_res_blocks,
hop_length=hp.hop_length,
sample_rate=hp.sample_rate,
mode=hp.voc_mode
)
# Hardware Optimization: Prefer CUDA for high-performance generation
if torch.cuda.is_available():
_model = _model.cuda()
_device = torch.device('cuda')
else:
_device = torch.device('cpu')
if verbose:
print("Loading model weights from: %s" % weights_fpath)
checkpoint = torch.load(weights_fpath, map_location=_device, weights_only=False)
_model.load_state_dict(checkpoint['model_state'])
_model.eval()
def is_loaded():
"""Status Check: Verifies if the vocoder is initialized in memory."""
return _model is not None
def infer_waveform(mel, normalize=True, batched=True, target=8000, overlap=800, progress_callback=None):
"""
Waveform Synthesis Phase:
Transforms a Mel-Spectrogram into a time-domain waveform using neural vocoding.
:param mel: Mel-Spectrogram input (numpy array)
:param normalize: Whether to scale the input spectrogram
:param batched: Use parallel generation for speed
:param target: Chunk size for batched synthesis
:param overlap: Samples used for smooth blending between chunks
:return: Synthesized waveform (numpy array)
"""
if _model is None:
raise Exception("Operational Error: Wave-RNN must be loaded before inference.")
if normalize:
mel = mel / hp.mel_max_abs_value
mel = torch.from_numpy(mel[None, ...])
wav = _model.generate(mel, batched, target, overlap, hp.mu_law, progress_callback)
return wav
|