nnfp/nn_fingerprint.py

# Copyright 2021 Zilliz. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import logging
import warnings

import os
from pathlib import Path
from typing import List

import torch
import numpy
import resampy
import onnxruntime

from towhee.operator.base import NNOperator
from towhee import register
from towhee.types.audio_frame import AudioFrame
from towhee.models.nnfp import NNFp
from towhee.models.utils.audio_preprocess import preprocess_wav, MelSpec

from .configs import default_params

warnings.filterwarnings('ignore')
log = logging.getLogger()


@register(output_schema=['vecs'])
class NNFingerprint(NNOperator):
    """
    Audio embedding operator using Neural Network Fingerprint
    """

    def __init__(self,
                 params: dict = None,
                 model_path: str = None,
                 framework: str = 'pytorch',
                 ):
        super().__init__(framework=framework)
        self.device = 'cuda' if torch.cuda.is_available() else 'cpu'
        if params is None:
            self.params = default_params
        else:
            self.params = params

        log.info('Loading model...')
        if model_path is None:
            path = str(Path(__file__).parent)
            model_path = os.path.join(path, 'saved_model', 'pfann_fma_m.pt')
        if model_path.endswith('.onnx'):
            log.warning('Using onnx.')
            self.model = onnxruntime.InferenceSession(model_path)
        else:
            state_dict = torch.load(model_path, map_location=self.device)
            if isinstance(state_dict, torch.nn.Module):
                self.model = state_dict
            else:
                dim = self.params['dim']
                h = self.params['h']
                u = self.params['u']
                f_bin = self.params['n_mels']
                n_seg = int(self.params['segment_size'] * self.params['sample_rate'])
                t = (n_seg + self.params['hop_length'] - 1) // self.params['hop_length']
                log.info('Creating model with parameters...')
                self.model = NNFp(
                    dim=dim, h=h, u=u,
                    in_f=f_bin, in_t=t,
                    fuller=self.params['fuller'],
                    activation=self.params['activation']
                ).to(self.device)
                self.model.load_state_dict(state_dict)
            self.model.eval()
        log.info('Model is loaded.')

    def __call__(self, data: List[AudioFrame]) -> numpy.ndarray:
        audio_tensors = self.preprocess(data).to(self.device)
        # print(audio_tensors.shape)
        if isinstance(self.model, onnxruntime.InferenceSession):
            audio_numpy = audio_tensors.detach().cpu().numpy() if audio_tensors.requires_grad \
                else audio_tensors.cpu().numpy()
            ort_inputs = {self.model.get_inputs()[0].name: audio_numpy}
            outs = self.model.run(None, ort_inputs)[0]
        else:
            features = self.model(audio_tensors)
            outs = features.detach().cpu().numpy()
        return outs

    def preprocess(self, frames: List[AudioFrame]):
        sr = frames[0].sample_rate
        layout = frames[0].layout
        if layout == 'stereo':
            frames = [frame.reshape(-1, 2) for frame in frames]
            audio = numpy.vstack(frames).transpose()
        else:
            audio = numpy.hstack(frames)
            audio = audio[None, :]
        audio = self.int2float(audio)

        if sr != self.params['sample_rate']:
            audio = resampy.resample(audio, sr, self.params['sample_rate'])

        wav = preprocess_wav(audio,
                             segment_size=int(self.params['sample_rate'] * self.params['segment_size']),
                             hop_size=int(self.params['sample_rate'] * self.params['hop_size']),
                             frame_shift_mul=self.params['frame_shift_mul']).to(self.device)
        wav = wav.to(torch.float32)
        mel = MelSpec(sample_rate=self.params['sample_rate'],
                      window_length=self.params['window_length'],
                      hop_length=self.params['hop_length'],
                      f_min=self.params['f_min'],
                      f_max=self.params['f_max'],
                      n_mels=self.params['n_mels'],
                      naf_mode=self.params['naf_mode'],
                      mel_log=self.params['mel_log'],
                      spec_norm=self.params['spec_norm']).to(self.device)
        wav = mel(wav)
        return wav

    @staticmethod
    def int2float(wav: numpy.ndarray, dtype: str = 'float64'):
        """
        Convert audio imgs from int to float.
        The input dtype must be integers.
        The output dtype is controlled by the parameter `dtype`, defaults to 'float64'.

        The code is inspired by https://github.com/mgeier/python-audio/blob/master/audio-files/utility.py
        """
        dtype = numpy.dtype(dtype)
        assert dtype.kind == 'f'

        if wav.dtype.kind in 'iu':
            ii = numpy.iinfo(wav.dtype)
            abs_max = 2 ** (ii.bits - 1)
            offset = ii.min + abs_max
            return (wav.astype(dtype) - offset) / abs_max
        else:
            log.warning('Converting float dtype from %s to %s.', wav.dtype, dtype)
            return wav.astype(dtype)

    def save_model(self, format: str = 'pytorch', path: str = 'default'):
        if path == 'default':
            path = str(Path(__file__).parent)
            path = os.path.join(path, 'saved', format)
            os.makedirs(path, exist_ok=True)
            name = 'nnfp'
            path = os.path.join(path, name)
        dummy_input = torch.rand(
            (1,) + (self.params['n_mels'], self.params['u'])
        ).to(self.device)
        if format == 'pytorch':
            path = path + '.pt'
            torch.save(self.model, path)
        elif format == 'torchscript':
            path = path + '.pt'
            try:
                try:
                    jit_model = torch.jit.script(self.model)
                except Exception:
                    log.warning(
                        'Failed to directly export as torchscript.'
                        'Using dummy input in shape of %s now.', dummy_input.shape)
                    jit_model = torch.jit.trace(self.model, dummy_input, strict=False)
                torch.jit.save(jit_model, path)
            except Exception as e:
                log.error(f'Fail to save as torchscript: {e}.')
                raise RuntimeError(f'Fail to save as torchscript: {e}.')
        elif format == 'onnx':
            path = path + '.onnx'
            try:
                torch.onnx.export(self.model,
                                  dummy_input,
                                  path,
                                  input_names=['input'],
                                  output_names=['output'],
                                  opset_version=12,
                                  do_constant_folding=True,
                                  dynamic_axes={'input': {0: 'batch_size'},
                                                'output': {0: 'batch_size'}
                                                }
                                  )
            except Exception as e:
                log.error(f'Fail to save as onnx: {e}.')
                raise RuntimeError(f'Fail to save as onnx: {e}.')
            # todo: elif format == 'tensorrt':
        else:
            log.error(f'Unsupported format "{format}".')

    def input_schema(self):
        return [(AudioFrame, (1024,))]

    def output_schema(self):
        return [(numpy.ndarray, (-1, self.params['dim']))]