diart 0.4.0

Speaker diarization in real time

Installation | Stream audio | Add your model | Tune hyper-parameters | Build pipelines
Research | Citation | Reproducibility

Installation

1. Create environment:

conda create -n diart python=3.8
conda activate diart

2. Install PortAudio and soundfile:

conda install portaudio
conda install pysoundfile -c conda-forge

3. Install PyTorch

4. Install pyannote.audio 2.0 (currently in development)

pip install git+https://github.com/pyannote/pyannote-audio.git@develop#egg=pyannote-audio

Note: starting from version 0.4, installing pyannote.audio is mandatory to run the default system or to use pyannote-based models. In any other case, this step can be ignored.

5. Install diart:

pip install diart

Stream audio

From the command line

A recorded conversation:

diart.stream /path/to/audio.wav

A live conversation:

diart.stream microphone

See diart.stream -h for more options.

From python

Run a real-time speaker diarization pipeline over an audio stream with RealTimeInference:

from diart.sources import MicrophoneAudioSource
from diart.inference import RealTimeInference
from diart.pipelines import OnlineSpeakerDiarization, PipelineConfig

config = PipelineConfig()  # Default parameters
pipeline = OnlineSpeakerDiarization(config)
audio_source = MicrophoneAudioSource(config.sample_rate)
inference = RealTimeInference("/output/path", do_plot=True)
inference(pipeline, audio_source)

For faster inference and evaluation on a dataset we recommend to use Benchmark instead (see our notes on reproducibility).

Add your model

Third-party segmentation and embedding models can be integrated seamlessly by subclassing SegmentationModel and EmbeddingModel:

import torch
from typing import Optional
from diart.models import EmbeddingModel
from diart.pipelines import PipelineConfig, OnlineSpeakerDiarization
from diart.sources import MicrophoneAudioSource
from diart.inference import RealTimeInference

class MyEmbeddingModel(EmbeddingModel):
    def __init__(self):
        super().__init__()
        self.my_pretrained_model = load("my_model.ckpt")
    
    def __call__(
        self,
        waveform: torch.Tensor,
        weights: Optional[torch.Tensor] = None
    ) -> torch.Tensor:
        return self.my_pretrained_model(waveform, weights)

config = PipelineConfig(embedding=MyEmbeddingModel())
pipeline = OnlineSpeakerDiarization(config)
mic = MicrophoneAudioSource(config.sample_rate)
inference = RealTimeInference("/out/dir")
inference(pipeline, mic)

Tune hyper-parameters

Diart implements a hyper-parameter optimizer based on optuna that allows you to tune any pipeline to any dataset.

From the command line

diart.tune /wav/dir --reference /rttm/dir --output /out/dir

See diart.tune -h for more options.

From python

from diart.optim import Optimizer, TauActive, RhoUpdate, DeltaNew
from diart.pipelines import PipelineConfig
from diart.inference import Benchmark

# Benchmark runs and evaluates the pipeline on a dataset
benchmark = Benchmark("/wav/dir", "/rttm/dir", "/out/dir/tmp", show_report=False)
# Base configuration for the pipeline we're going to tune
base_config = PipelineConfig()
# Hyper-parameters to optimize
hparams = [TauActive, RhoUpdate, DeltaNew]
# Optimizer implements the optimization loop
optimizer = Optimizer(benchmark, base_config, hparams, "/out/dir")
# Run optimization
optimizer.optimize(num_iter=100, show_progress=True)

This will use /out/dir/tmp as a working directory and write results to an sqlite database in /out/dir.

Distributed optimization

For bigger datasets, it is sometimes more convenient to run multiple optimization processes in parallel. To do this, create a study on a recommended DBMS (e.g. MySQL or PostgreSQL) making sure that the study and database names match:

mysql -u root -e "CREATE DATABASE IF NOT EXISTS example"
optuna create-study --study-name "example" --storage "mysql://root@localhost/example"

Then you can run multiple identical optimizers pointing to the database:

diart.tune /wav/dir --reference /rttm/dir --output /out/dir --storage mysql://root@localhost/example

If you are using the python API, make sure that worker directories are different to avoid concurrency issues:

from diart.optim import Optimizer
from diart.inference import Benchmark
from optuna.samplers import TPESampler
import optuna

ID = 0  # Worker identifier
base_config, hparams = ...
benchmark = Benchmark("/wav/dir", "/rttm/dir", f"/out/dir/worker-{ID}", show_report=False)
study = optuna.load_study("example", "mysql://root@localhost/example", TPESampler())
optimizer = Optimizer(benchmark, base_config, hparams, study)
optimizer.optimize(num_iter=100, show_progress=True)

Build pipelines

For a more advanced usage, diart also provides building blocks that can be combined to create your own pipeline. Streaming is powered by RxPY, but the blocks module is completely independent and can be used separately.

Example

Obtain overlap-aware speaker embeddings from a microphone stream:

import rx.operators as ops
import diart.operators as dops
from diart.sources import MicrophoneAudioSource
from diart.blocks import SpeakerSegmentation, OverlapAwareSpeakerEmbedding

segmentation = SpeakerSegmentation.from_pyannote("pyannote/segmentation")
embedding = OverlapAwareSpeakerEmbedding.from_pyannote("pyannote/embedding")
sample_rate = segmentation.model.get_sample_rate()
mic = MicrophoneAudioSource(sample_rate)

stream = mic.stream.pipe(
    # Reformat stream to 5s duration and 500ms shift
    dops.regularize_audio_stream(sample_rate),
    ops.map(lambda wav: (wav, segmentation(wav))),
    ops.starmap(embedding)
).subscribe(on_next=lambda emb: print(emb.shape))

mic.read()

Output:

torch.Size([4, 512])
torch.Size([4, 512])
torch.Size([4, 512])
...

Powered by research

Diart is the official implementation of the paper Overlap-aware low-latency online speaker diarization based on end-to-end local segmentation by Juan Manuel Coria, Hervé Bredin, Sahar Ghannay and Sophie Rosset.

We propose to address online speaker diarization as a combination of incremental clustering and local diarization applied to a rolling buffer updated every 500ms. Every single step of the proposed pipeline is designed to take full advantage of the strong ability of a recently proposed end-to-end overlap-aware segmentation to detect and separate overlapping speakers. In particular, we propose a modified version of the statistics pooling layer (initially introduced in the x-vector architecture) to give less weight to frames where the segmentation model predicts simultaneous speakers. Furthermore, we derive cannot-link constraints from the initial segmentation step to prevent two local speakers from being wrongfully merged during the incremental clustering step. Finally, we show how the latency of the proposed approach can be adjusted between 500ms and 5s to match the requirements of a particular use case, and we provide a systematic analysis of the influence of latency on the overall performance (on AMI, DIHARD and VoxConverse).

Citation

If you found diart useful, please make sure to cite our paper:

@inproceedings{diart,  
  author={Coria, Juan M. and Bredin, Hervé and Ghannay, Sahar and Rosset, Sophie},  
  booktitle={2021 IEEE Automatic Speech Recognition and Understanding Workshop (ASRU)},   
  title={Overlap-Aware Low-Latency Online Speaker Diarization Based on End-to-End Local Segmentation}, 
  year={2021},
  pages={1139-1146},
  doi={10.1109/ASRU51503.2021.9688044},
}

Reproducibility

Diart aims to be lightweight and capable of real-time streaming in practical scenarios. Its performance is very close to what is reported in the paper (and sometimes even a bit better).

To obtain the best results, make sure to use the following hyper-parameters:

Dataset	latency	tau	rho	delta
DIHARD III	any	0.555	0.422	1.517
AMI	any	0.507	0.006	1.057
VoxConverse	any	0.576	0.915	0.648
DIHARD II	1s	0.619	0.326	0.997
DIHARD II	5s	0.555	0.422	1.517

diart.benchmark and diart.inference.Benchmark can quickly run and evaluate the pipeline, and even measure its real-time latency. For instance, for a DIHARD III configuration:

diart.benchmark /wav/dir --reference /rttm/dir --tau=0.555 --rho=0.422 --delta=1.517 --output /out/dir

or using the inference API:

from diart.inference import Benchmark
from diart.pipelines import OnlineSpeakerDiarization, PipelineConfig

config = PipelineConfig(
    step=0.5,
    latency=0.5,
    tau_active=0.555,
    rho_update=0.422,
    delta_new=1.517
)
pipeline = OnlineSpeakerDiarization(config)
benchmark = Benchmark("/wav/dir", "/rttm/dir", "/out/dir")

benchmark(pipeline)

This runs a faster inference by pre-calculating model outputs in batches. See diart.benchmark -h for more options.

For convenience and to facilitate future comparisons, we also provide the expected outputs of the paper implementation in RTTM format for every entry of Table 1 and Figure 5. This includes the VBx offline topline as well as our proposed online approach with latencies 500ms, 1s, 2s, 3s, 4s, and 5s.

License

MIT License

Copyright (c) 2021 Université Paris-Saclay
Copyright (c) 2021 CNRS

Permission is hereby granted, free of charge, to any person obtaining a copy
of this software and associated documentation files (the "Software"), to deal
in the Software without restriction, including without limitation the rights
to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
copies of the Software, and to permit persons to whom the Software is
furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all
copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
SOFTWARE.

Logo generated by DesignEvo free logo designer