tools/asr_webapp/model_api.py

# Copyright (c) 2021, NVIDIA CORPORATION.  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 contextlib
import glob
import os

import torch

import nemo.collections.asr as nemo_asr
from nemo.utils import logging, model_utils

# setup AMP (optional)
if torch.cuda.is_available() and hasattr(torch.cuda, 'amp') and hasattr(torch.cuda.amp, 'autocast'):
    logging.info("AMP enabled!\n")
    autocast = torch.cuda.amp.autocast
else:

    @contextlib.contextmanager
    def autocast():
        yield


MODEL_CACHE = {}

# Special tags for fallbacks / user notifications
TAG_ERROR_DURING_TRANSCRIPTION = "<ERROR_DURING_TRANSCRIPTION>"


def get_model_names():
    # Populate local copy of models
    local_model_paths = glob.glob(os.path.join('models', "**", "*.nemo"), recursive=True)
    local_model_names = list(sorted([os.path.basename(path) for path in local_model_paths]))

    # Populate with pretrained nemo checkpoint list
    nemo_model_names = set()
    for model_info in nemo_asr.models.ASRModel.list_available_models():
        for superclass in model_info.class_.mro():
            if 'CTC' in superclass.__name__ or 'RNNT' in superclass.__name__:
                if 'align' in model_info.pretrained_model_name:
                    continue

                nemo_model_names.add(model_info.pretrained_model_name)
    nemo_model_names = list(sorted(nemo_model_names))
    return nemo_model_names, local_model_names


def initialize_model(model_name):
    # load model
    if model_name not in MODEL_CACHE:
        if '.nemo' in model_name:
            # use local model
            model_name_no_ext = os.path.splitext(model_name)[0]
            model_path = os.path.join('models', model_name_no_ext, model_name)

            # Extract config
            model_cfg = nemo_asr.models.ASRModel.restore_from(restore_path=model_path, return_config=True)
            classpath = model_cfg.target  # original class path
            imported_class = model_utils.import_class_by_path(classpath)  # type: ASRModel
            logging.info(f"Restoring local model : {imported_class.__name__}")

            # load model from checkpoint
            model = imported_class.restore_from(restore_path=model_path, map_location='cpu')  # type: ASRModel

        else:
            # use pretrained model
            model = nemo_asr.models.ASRModel.from_pretrained(model_name, map_location='cpu')

        model.freeze()

        # cache model
        MODEL_CACHE[model_name] = model

    model = MODEL_CACHE[model_name]
    return model


def transcribe_all(filepaths, model_name, use_gpu_if_available=True):
    # instantiate model
    if model_name in MODEL_CACHE:
        model = MODEL_CACHE[model_name]
    else:
        model = initialize_model(model_name)

    if torch.cuda.is_available() and use_gpu_if_available:
        model = model.cuda()

    # transcribe audio
    logging.info("Begin transcribing audio...")
    try:
        with autocast():
            with torch.no_grad():
                transcriptions = model.transcribe(filepaths, batch_size=32)

    except RuntimeError:
        # Purge the cache to clear some memory
        MODEL_CACHE.clear()

        logging.info("Ran out of memory on device - performing inference on CPU for now")

        try:
            model = model.cpu()

            with torch.no_grad():
                transcriptions = model.transcribe(filepaths, batch_size=32)

        except Exception as e:
            logging.info(f"Exception {e} occured while attemting to transcribe audio. Returning error message")
            return TAG_ERROR_DURING_TRANSCRIPTION

    logging.info(f"Finished transcribing {len(filepaths)} files !")

    # If RNNT models transcribe, they return a tuple (greedy, beam_scores)
    if type(transcriptions[0]) == list and len(transcriptions) == 2:
        # get greedy transcriptions only
        transcriptions = transcriptions[0]

    # Force onto CPU
    model = model.cpu()
    MODEL_CACHE[model_name] = model

    return transcriptions