{"id":13784457,"url":"https://github.com/ksw0306/FloWaveNet","last_synced_at":"2025-05-11T19:32:59.706Z","repository":{"id":111961681,"uuid":"155968058","full_name":"ksw0306/FloWaveNet","owner":"ksw0306","description":"A Pytorch implementation of \"FloWaveNet: A Generative Flow for Raw Audio\"","archived":false,"fork":false,"pushed_at":"2019-04-23T13:57:02.000Z","size":59,"stargazers_count":490,"open_issues_count":4,"forks_count":109,"subscribers_count":42,"default_branch":"master","last_synced_at":"2024-11-17T20:48:32.659Z","etag":null,"topics":["clarinet","generative-flow","glow","pytorch","wavenet"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/ksw0306.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null}},"created_at":"2018-11-03T09:21:42.000Z","updated_at":"2024-09-06T05:52:59.000Z","dependencies_parsed_at":"2023-07-08T23:15:29.717Z","dependency_job_id":null,"html_url":"https://github.com/ksw0306/FloWaveNet","commit_stats":null,"previous_names":[],"tags_count":0,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ksw0306%2FFloWaveNet","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ksw0306%2FFloWaveNet/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ksw0306%2FFloWaveNet/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/ksw0306%2FFloWaveNet/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/ksw0306","download_url":"https://codeload.github.com/ksw0306/FloWaveNet/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":253621329,"owners_count":21937505,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["clarinet","generative-flow","glow","pytorch","wavenet"],"created_at":"2024-08-03T19:00:44.475Z","updated_at":"2025-05-11T19:32:59.437Z","avatar_url":"https://github.com/ksw0306.png","language":"Python","funding_links":[],"categories":["Python","📝 Publications \u003csmall\u003e(60)\u003c/small\u003e"],"sub_categories":[],"readme":"# FloWaveNet : A Generative Flow for Raw Audio \n\nThis is a PyTorch implementation of our work [\"FloWaveNet : A Generative Flow for Raw Audio\".](https://arxiv.org/abs/1811.02155) (We'll update soon.)\n\n\u003cimg src=\"png/model.png\"\u003e\n\nFor a purpose of parallel sampling, we propose FloWaveNet, a flow-based generative model for raw audio synthesis.\nFloWaveNet can generate audio samples as fast as ClariNet and Parallel WaveNet, while the training procedure is really easy and stable with a single-stage pipeline. Our generated audio samples are available at [https://ksw0306.github.io/flowavenet-demo/](https://ksw0306.github.io/flowavenet-demo/). Also, our implementation of ClariNet (Gaussian WaveNet and Gaussian IAF) is available at [https://github.com/ksw0306/ClariNet](https://github.com/ksw0306/ClariNet)\n\n\n# Requirements\n\n- PyTorch 0.4.1\n- Python 3.6\n- Librosa\n\n# Examples\n\n#### Step 1. Download Dataset\n\n- LJSpeech : [https://keithito.com/LJ-Speech-Dataset/](https://keithito.com/LJ-Speech-Dataset/)\n\n#### Step 2. Preprocessing (Preparing Mel Spectrogram)\n\n`python preprocessing.py --in_dir ljspeech --out_dir DATASETS/ljspeech`\n\n#### Step 3. Train\n\n##### Single-GPU training\n\n`python train.py --model_name flowavenet --batch_size 2 --n_block 8 --n_flow 6 --n_layer 2 --block_per_split 4`\n\n##### Multi-GPU training\n\n`python train.py --model_name flowavenet --batch_size 8 --n_block 8 --n_flow 6 --n_layer 2 --block_per_split 4 --num_gpu 4`\n\n\nNVIDIA TITAN V (12GB VRAM) : batch size 2 per GPU\n\nNVIDIA Tesla V100 (32GB VRAM) : batch size 8 per GPU\n\n\n#### Step 4. Synthesize\n\n`--load_step CHECKPOINT` : the # of the pre-trained model's global training step (also depicted in the trained weight file)\n\n`--temp`: Temperature (standard deviation) value implemented as z ~ N(0, 1 * TEMPERATURE^2)\n\nex) `python synthesize.py --model_name flowavenet --n_block 8 --n_flow 6 --n_layer 2 --load_step 100000 --temp 0.8 --num_samples 10 --block_per_split 4`\n\n\n\n# Sample Link\n\nSample Link : [https://ksw0306.github.io/flowavenet-demo/](https://ksw0306.github.io/flowavenet-demo/)\n\nOur implementation of ClariNet (Gaussian WaveNet, Gaussian IAF) : [https://github.com/ksw0306/ClariNet](https://github.com/ksw0306/ClariNet)\n\n- Results 1 : Model Comparisons (WaveNet (MoL, Gaussian), ClariNet and FloWaveNet)\n\n- Results 2 : Temperature effect on Audio Quality Trade-off (Temperature T : 0.0 ~ 1.0, Model : FloWaveNet)\n\n- Results 3 : Analysis of ClariNet Loss Terms (Loss functions : 1. Only KL 2. KL + Frame 3. Only Frame)\n\n- Results 4 : Causality of WaveNet Dilated Convolutions (FloWaveNet : Non-causal WaveNet Affine Coupling Layers, FloWaveNet_causal : Causal WaveNet Affine Coupling Layers)\n\n\n# Reference\n\n- WaveNet vocoder : [https://github.com/r9y9/wavenet_vocoder](https://github.com/r9y9/wavenet_vocoder)\n- glow-pytorch : [https://github.com/rosinality/glow-pytorch](https://github.com/rosinality/glow-pytorch)\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fksw0306%2FFloWaveNet","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fksw0306%2FFloWaveNet","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fksw0306%2FFloWaveNet/lists"}