initial push

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.txt filter=lfs diff=lfs merge=lfs -text

best_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeac665e0b873dc01cd65dbfe387e522ef9b9dc05459b231d4f9ab495b965e76
+size 998061765

best_model_66651.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:eeac665e0b873dc01cd65dbfe387e522ef9b9dc05459b231d4f9ab495b965e76
+size 998061765

checkpoint_118000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5062823a325b0e295be9e136a8a8079861f92059acaea9afb890ba3111071407
+size 998061765

checkpoint_119000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5ef816e760bc1ba59e5f5d21a7d98fb5b3b6ab78e2709eb89ad2615b17b0da81
+size 998061765

checkpoint_120000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:acffaee30aaa0903c3e96e9b2f265057f5db7b4c2b0cea586bb476536b020172
+size 998061765

checkpoint_121000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:655e7d191e8da24fb8cc57166cee31c3ce9b473cb417469a75e69ab27d83b7d0
+size 998061765

checkpoint_122000.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:34d7ff228c0cec0c32b336f7b1b455015ec5adfd780837c40be32a3c1f6e5683
+size 998061765

config.json

@@ -0,0 +1,276 @@
+{
+    "output_path": "runs",
+    "logger_uri": null,
+    "run_name": "persian-tts-vits-grapheme-arman24-FINETUNED",
+    "project_name": null,
+    "run_description": "\ud83d\udc38Coqui trainer run.",
+    "print_step": 30,
+    "plot_step": 100,
+    "model_param_stats": false,
+    "wandb_entity": null,
+    "dashboard_logger": "tensorboard",
+    "save_on_interrupt": true,
+    "log_model_step": null,
+    "save_step": 1000,
+    "save_n_checkpoints": 5,
+    "save_checkpoints": true,
+    "save_all_best": false,
+    "save_best_after": 10000,
+    "target_loss": null,
+    "print_eval": true,
+    "test_delay_epochs": -1,
+    "run_eval": true,
+    "run_eval_steps": 1000,
+    "distributed_backend": "nccl",
+    "distributed_url": "tcp://localhost:54321",
+    "mixed_precision": false,
+    "precision": "fp16",
+    "epochs": 2000,
+    "batch_size": 1,
+    "eval_batch_size": 1,
+    "grad_clip": [
+        1000,
+        1000
+    ],
+    "scheduler_after_epoch": true,
+    "lr": 0.001,
+    "optimizer": "AdamW",
+    "optimizer_params": {
+        "betas": [
+            0.8,
+            0.99
+        ],
+        "eps": 1e-09,
+        "weight_decay": 0.01
+    },
+    "lr_scheduler": null,
+    "lr_scheduler_params": {},
+    "use_grad_scaler": false,
+    "allow_tf32": false,
+    "cudnn_enable": true,
+    "cudnn_deterministic": false,
+    "cudnn_benchmark": false,
+    "training_seed": 54321,
+    "model": "vits",
+    "num_loader_workers": 16,
+    "num_eval_loader_workers": 16,
+    "use_noise_augment": false,
+    "audio": {
+        "fft_size": 1024,
+        "sample_rate": 22050,
+        "win_length": 1024,
+        "hop_length": 256,
+        "num_mels": 80,
+        "mel_fmin": 0,
+        "mel_fmax": null
+    },
+    "use_phonemes": false,
+    "phonemizer": null,
+    "phoneme_language": null,
+    "compute_input_seq_cache": true,
+    "text_cleaner": "basic_cleaners",
+    "enable_eos_bos_chars": false,
+    "test_sentences_file": "",
+    "phoneme_cache_path": "cache/phoneme_cache_grapheme_azure-2",
+    "characters": {
+        "characters_class": "TTS.tts.models.vits.VitsCharacters",
+        "vocab_dict": null,
+        "pad": "<PAD>",
+        "eos": "<EOS>",
+        "bos": "<BOS>",
+        "blank": "<BLNK>",
+        "characters": "\u0621\u0627\u0628\u062a\u062b\u062c\u062d\u062e\u062f\u0630\u0631\u0632\u0633\u0634\u0635\u0636\u0637\u0638\u0639\u063a\u0641\u0642\u0644\u0645\u0646\u0647\u0648\u064a\u0650\u067e\u0686\u0698\u06a9\u06af\u06cc\u0622\u0623\u0624\u0625\u0626\u064b\u064e\u064f\u0651",
+        "punctuations": "!(),-.:;? \u0320\u060c\u061b\u061f\u200c<>\u066b",
+        "phonemes": "\u02c8\u02cc\u02d0\u02d1pbtd\u0288\u0256c\u025fk\u0261q\u0262\u0294\u0274\u014b\u0272\u0273n\u0271m\u0299r\u0280\u2c71\u027e\u027d\u0278\u03b2fv\u03b8\u00f0sz\u0283\u0292\u0282\u0290\u00e7\u029dx\u0263\u03c7\u0281\u0127\u0295h\u0266\u026c\u026e\u028b\u0279\u027bj\u0270l\u026d\u028e\u029faegiouwy\u026a\u028a\u0329\u00e6\u0251\u0254\u0259\u025a\u025b\u025d\u0268\u0303\u0289\u028c\u028d0123456789\"#$%*+/=ABCDEFGHIJKLMNOPRSTUVWXYZ[]^_{}\u06f1\u06f2\u06f3\u06f4\u06f5\u06f6\u06f7\u06f8\u06f9\u06f0",
+        "is_unique": true,
+        "is_sorted": true
+    },
+    "add_blank": true,
+    "batch_group_size": 0,
+    "loss_masking": null,
+    "min_audio_len": 1,
+    "max_audio_len": Infinity,
+    "min_text_len": 1,
+    "max_text_len": Infinity,
+    "compute_f0": false,
+    "compute_energy": false,
+    "compute_linear_spec": true,
+    "precompute_num_workers": 0,
+    "start_by_longest": false,
+    "shuffle": false,
+    "drop_last": false,
+    "datasets": [
+        {
+            "formatter": "mozilla",
+            "dataset_name": "",
+            "path": "./datasets/arman-24",
+            "meta_file_train": "metadata-hotfix.csv",
+            "ignored_speakers": null,
+            "language": "",
+            "phonemizer": "",
+            "meta_file_val": "",
+            "meta_file_attn_mask": ""
+        }
+    ],
+    "test_sentences": [
+        [
+            "\u0632\u06cc\u0646 \u0647\u0645\u0631\u0647\u0627\u0646 \u0633\u0633\u062a \u0639\u0646\u0627\u0635\u0631\u060c \u062f\u0644\u0645 \u06af\u0631\u0641\u062a."
+        ],
+        [
+            "\u0628\u06cc\u0627 \u062a\u0627 \u06af\u0644 \u0628\u0631\u0627\u0641\u0634\u0627\u0646\u06cc\u0645 \u0648 \u0645\u06cc \u062f\u0631 \u0633\u0627\u063a\u0631 \u0627\u0646\u062f\u0627\u0632\u06cc\u0645."
+        ],
+        [
+            "\u0628\u0646\u06cc \u0622\u062f\u0645 \u0627\u0639\u0636\u0627\u06cc \u06cc\u06a9 \u067e\u06cc\u06a9\u0631\u0646\u062f, \u06a9\u0647 \u062f\u0631 \u0622\u0641\u0631\u06cc\u0646\u0634 \u0632 \u06cc\u06a9 \u06af\u0648\u0647\u0631\u0646\u062f."
+        ],
+        [
+            "\u0633\u0647\u0627\u0645 \u0632\u0646\u062f\u06af\u06cc \u0628\u0647 10 \u062f\u0631\u0635\u062f \u0648 \u0633\u0647\u0627\u0645 \u0628\u06cc\u062a\u06a9\u0648\u06cc\u0646 \u06af\u0648\u06af\u0644 \u0628\u0647 33 \u062f\u0631\u0635\u062f \u0627\u0641\u0632\u0627\u06cc\u0634 \u06cc\u0627\u0641\u062a."
+        ],
+        [
+            "\u0645\u0646 \u0628\u0648\u062f\u0645 \u0648 \u0622\u0628\u062c\u06cc \u0641\u0648\u062a\u06cc\u0646\u0627\u060c \u0648 \u062d\u0627\u0644\u0627 \u0631\u067e\u062a\u06cc \u067e\u062a\u06cc\u0646\u0627. \u0627\u06cc\u0646 \u0634\u0639\u0631 \u06cc\u06a9\u06cc \u0627\u0632 \u0627\u0634\u0639\u0627\u0631 \u0645\u0639\u0631\u0648\u0641 \u0631\u0648 \u062d\u0648\u0636\u06cc \u0627\u0633\u062a \u06a9\u0647 \u062f\u0631 \u06a9\u0648\u0686\u0647 \u0628\u0627\u0632\u0627\u0631 \u062a\u0647\u0631\u0627\u0646 \u0632\u0645\u0632\u0645\u0647 \u0645\u06cc \u0634\u062f\u0647 \u0627\u0633\u062a."
+        ],
+        [
+            "\u06cc\u0647 \u062f\u0648 \u062f\u0642\u0647 \u0647\u0645 \u0628\u0647 \u062d\u0631\u0641\u0645 \u06af\u0648\u0634 \u06a9\u0646\u060c \u0646\u06af\u0648 \u0646\u06af\u0648\u0634\u06cc\u062f\u0645 \u0648 \u0646\u062d\u0631\u0641\u06cc\u062f\u06cc."
+        ],
+        [
+            "\u062f\u0627\u0633\u062a\u0627\u0646 \u0628\u0627 \u062a\u0648\u0635\u06cc\u0641 \u0637\u0648\u0641\u0627\u0646\u200c\u0647\u0627\u06cc \u0634\u062f\u06cc\u062f \u0622\u063a\u0627\u0632 \u0645\u06cc\u200c\u0634\u0648\u062f\u061b \u0637\u0648\u0641\u0627\u0646\u200c\u0647\u0627\u06cc\u06cc \u06a9\u0647 \u0645\u0632\u0631\u0639\u0647\u200c\u0647\u0627 \u0631\u0627 \u0627\u0632 \u0628\u06cc\u0646 \u0645\u06cc\u200c\u0628\u0631\u062f \u0648 \u0645\u062d\u0635\u0648\u0644\u0627\u062a \u0631\u0627 \u0632\u06cc\u0631 \u0634\u0646 \u062f\u0641\u0646 \u0645\u06cc\u200c\u06a9\u0646\u062f\u061b \u0645\u062d\u0635\u0648\u0644\u0627\u062a\u06cc \u06a9\u0647 \u0632\u0646\u062f\u06af\u06cc \u0627\u0641\u0631\u0627\u062f \u0628\u0633\u06cc\u0627\u0631\u06cc \u0628\u0647 \u0622\u0646 \u0648\u0627\u0628\u0633\u062a\u0647 \u0627\u0633\u062a."
+        ]
+    ],
+    "eval_split_max_size": null,
+    "eval_split_size": 0.01,
+    "use_speaker_weighted_sampler": false,
+    "speaker_weighted_sampler_alpha": 1.0,
+    "use_language_weighted_sampler": false,
+    "language_weighted_sampler_alpha": 1.0,
+    "use_length_weighted_sampler": false,
+    "length_weighted_sampler_alpha": 1.0,
+    "model_args": {
+        "num_chars": 209,
+        "out_channels": 513,
+        "spec_segment_size": 32,
+        "hidden_channels": 192,
+        "hidden_channels_ffn_text_encoder": 768,
+        "num_heads_text_encoder": 2,
+        "num_layers_text_encoder": 6,
+        "kernel_size_text_encoder": 3,
+        "dropout_p_text_encoder": 0.1,
+        "dropout_p_duration_predictor": 0.5,
+        "kernel_size_posterior_encoder": 5,
+        "dilation_rate_posterior_encoder": 1,
+        "num_layers_posterior_encoder": 16,
+        "kernel_size_flow": 5,
+        "dilation_rate_flow": 1,
+        "num_layers_flow": 4,
+        "resblock_type_decoder": "1",
+        "resblock_kernel_sizes_decoder": [
+            3,
+            7,
+            11
+        ],
+        "resblock_dilation_sizes_decoder": [
+            [
+                1,
+                3,
+                5
+            ],
+            [
+                1,
+                3,
+                5
+            ],
+            [
+                1,
+                3,
+                5
+            ]
+        ],
+        "upsample_rates_decoder": [
+            8,
+            8,
+            2,
+            2
+        ],
+        "upsample_initial_channel_decoder": 512,
+        "upsample_kernel_sizes_decoder": [
+            16,
+            16,
+            4,
+            4
+        ],
+        "periods_multi_period_discriminator": [
+            2,
+            3,
+            5,
+            7,
+            11
+        ],
+        "use_sdp": true,
+        "noise_scale": 1.0,
+        "inference_noise_scale": 0.667,
+        "length_scale": 1,
+        "noise_scale_dp": 1.0,
+        "inference_noise_scale_dp": 1.0,
+        "max_inference_len": null,
+        "init_discriminator": true,
+        "use_spectral_norm_disriminator": false,
+        "use_speaker_embedding": false,
+        "num_speakers": 0,
+        "speakers_file": null,
+        "d_vector_file": null,
+        "speaker_embedding_channels": 256,
+        "use_d_vector_file": false,
+        "d_vector_dim": 0,
+        "detach_dp_input": true,
+        "use_language_embedding": false,
+        "embedded_language_dim": 4,
+        "num_languages": 0,
+        "language_ids_file": null,
+        "use_speaker_encoder_as_loss": false,
+        "speaker_encoder_config_path": "",
+        "speaker_encoder_model_path": "",
+        "condition_dp_on_speaker": true,
+        "freeze_encoder": false,
+        "freeze_DP": false,
+        "freeze_PE": false,
+        "freeze_flow_decoder": false,
+        "freeze_waveform_decoder": false,
+        "encoder_sample_rate": null,
+        "interpolate_z": true,
+        "reinit_DP": false,
+        "reinit_text_encoder": false
+    },
+    "lr_gen": 0.0002,
+    "lr_disc": 0.0002,
+    "lr_scheduler_gen": "ExponentialLR",
+    "lr_scheduler_gen_params": {
+        "gamma": 0.999875,
+        "last_epoch": -1
+    },
+    "lr_scheduler_disc": "ExponentialLR",
+    "lr_scheduler_disc_params": {
+        "gamma": 0.999875,
+        "last_epoch": -1
+    },
+    "kl_loss_alpha": 1.0,
+    "disc_loss_alpha": 1.0,
+    "gen_loss_alpha": 1.0,
+    "feat_loss_alpha": 1.0,
+    "mel_loss_alpha": 45.0,
+    "dur_loss_alpha": 1.0,
+    "speaker_encoder_loss_alpha": 1.0,
+    "return_wav": true,
+    "use_weighted_sampler": false,
+    "weighted_sampler_attrs": {},
+    "weighted_sampler_multipliers": {},
+    "r": 1,
+    "num_speakers": 0,
+    "use_speaker_embedding": false,
+    "speakers_file": null,
+    "speaker_embedding_channels": 256,
+    "language_ids_file": null,
+    "use_language_embedding": false,
+    "use_d_vector_file": false,
+    "d_vector_file": null,
+    "d_vector_dim": 0,
+    "restore_path": "/home/mhr/runs/persian-tts-vits-grapheme-arman24-October-18-2023_07+02AM-0000000-FINETUNE/checkpoint_53000.pth",
+    "github_branch": "inside_docker"
+}

events.out.tfevents.1697758857.sn4622110887.454266.0

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:23cd4b11f52a4f7c1995c71d9d3a3dea687e51abfe5f10625e5c706a7ffa4b1b
+size 2905602015

recipe-train-vits-10-grphm-arman24.py

@@ -0,0 +1,164 @@
+import os
+# os.environ["CUDA_VISIBLE_DEVICES"] = "7"
+
+import wandb
+from trainer import Trainer, TrainerArgs
+
+from TTS.tts.configs.shared_configs import BaseDatasetConfig , CharactersConfig
+from TTS.config.shared_configs import BaseAudioConfig
+from TTS.tts.configs.vits_config import VitsConfig
+from TTS.tts.datasets import load_tts_samples
+from TTS.tts.models.vits import Vits, VitsAudioConfig, VitsArgs
+from TTS.tts.utils.text.tokenizer import TTSTokenizer
+from TTS.utils.audio import AudioProcessor
+from TTS.tts.utils.speakers import SpeakerManager
+
+ # Start a wandb run with `sync_tensorboard=True`
+if wandb.run is None:
+    wandb.init(project="persian-tts-vits-grapheme-arman24-FineTuned", group="GPUx8 accel mixed fp16 1x32", sync_tensorboard=True)
+
+# output_path = os.path.dirname(os.path.abspath(__file__))
+# output_path = output_path + '/notebook_files/runs'
+# output_path = wandb.run.dir  ### PROBABLY better for notebook
+output_path = "runs"
+
+print("output path is:")
+print(output_path)
+
+cache_path = "cache"
+dataset_config = BaseDatasetConfig(
+    formatter="mozilla", meta_file_train="metadata-hotfix.csv", path="./datasets/arman-24"
+)
+
+character_config=CharactersConfig(
+  characters='ءابتثجحخدذرزسشصضطظعغفقلمنهويِپچژکگیآأؤإئًَُّ',
+#   characters="!¡'(),-.:;¿?ABCDEFGHIJKLMNOPRSTUVWXYZabcdefghijklmnopqrstuvwxyzáçèéêëìíîïñòóôöùúûü«°±µ»$%&‘’‚“`”„",
+  punctuations='!(),-.:;? ̠،؛؟‌<>٫',
+  phonemes='ˈˌːˑpbtdʈɖcɟkɡqɢʔɴŋɲɳnɱmʙrʀⱱɾɽɸβfvθðszʃʒʂʐçʝxɣχʁħʕhɦɬɮʋɹɻjɰlɭʎʟaegiouwyɪʊ̩æɑɔəɚɛɝɨ̃ʉʌʍ0123456789"#$%*+/=ABCDEFGHIJKLMNOPRSTUVWXYZ[]^_{}۱۲۳۴۵۶۷۸۹۰',
+  pad="<PAD>",
+  eos="<EOS>",
+  bos="<BOS>",
+  blank="<BLNK>",
+  characters_class="TTS.tts.models.vits.VitsCharacters",
+  )
+
+# # From the coqui multilinguL recipes, will try later
+# vitsArgs = VitsArgs(
+#     use_language_embedding=True,
+#     embedded_language_dim=1,
+#     use_speaker_embedding=True,
+#     use_sdp=False,
+# )
+
+audio_config = BaseAudioConfig(
+     sample_rate=22050,
+     do_trim_silence=True,
+     min_level_db=-1,
+    # do_sound_norm=True,
+     signal_norm=True,
+     clip_norm=True,
+     symmetric_norm=True,
+     max_norm = 0.9,
+     resample=True,
+     win_length=1024,
+     hop_length=256,
+     num_mels=80,
+     mel_fmin=0,
+     mel_fmax=None 
+ )
+
+vits_audio_config = VitsAudioConfig(
+    sample_rate=22050,
+#    do_sound_norm=True,
+    win_length=1024,
+    hop_length=256,
+    num_mels=80,
+    # do_trim_silence=True, #from hugging
+    mel_fmin=0,
+    mel_fmax=None
+)
+config = VitsConfig(
+    # model_args=vitsArgs,
+    audio=vits_audio_config, #from huggingface
+    run_name="persian-tts-vits-grapheme-arman24-FINETUNED",
+    batch_size=1,
+    batch_group_size=0,
+    eval_batch_size=1,
+    num_loader_workers=16,
+    num_eval_loader_workers=16,
+    run_eval=True,
+    run_eval_steps = 1000,
+    print_eval=True,
+    test_delay_epochs=-1,
+    epochs=2000,
+    save_step=1000,
+    text_cleaner="basic_cleaners", #from MH
+    use_phonemes=False,
+    # phonemizer='persian_mh', #from TTS github
+    # phoneme_language="fa",
+    characters=character_config, #test without as well
+    phoneme_cache_path=os.path.join(cache_path, "phoneme_cache_grapheme_azure-2"),
+    compute_input_seq_cache=True,
+    print_step=30,
+    mixed_precision=False, #from TTS - True causes error "Expected reduction dim"
+    test_sentences=[
+        ["زین همرهان سست عناصر، دلم گرفت."],
+        ["بیا تا گل برافشانیم و می در ساغر اندازیم."],
+        ["بنی آدم اعضای یک پیکرند, که در آفرینش ز یک گوهرند."],
+        ["سهام زندگی به 10 درصد و سهام بیتکوین گوگل به 33 درصد افزایش یافت."],
+        ["من بودم و آبجی فوتینا، و حالا رپتی پتینا. این شعر یکی از اشعار معروف رو حوضی است که در کوچه بازار تهران زمزمه می شده است." ],
+        ["یه دو دقه هم به حرفم گوش کن، نگو نگوشیدم و نحرفیدی."],
+        [ "داستان با توصیف طوفان‌های شدید آغاز می‌شود؛ طوفان‌هایی که مزرعه‌ها را از بین می‌برد و محصولات را زیر شن دفن می‌کند؛ محصولاتی که زندگی افراد بسیاری به آن وابسته است."]
+    ],
+    output_path=output_path,
+    datasets=[dataset_config],
+    # use_sdp=False,  ### used in multiligual recipes on coqui
+    # use_speaker_embedding=True
+)
+
+# INITIALIZE THE AUDIO PROCESSOR
+# Audio processor is used for feature extraction and audio I/O.
+# It mainly serves to the dataloader and the training loggers.
+ap = AudioProcessor.init_from_config(config)
+
+# INITIALIZE THE TOKENIZER
+# Tokenizer is used to convert text to sequences of token IDs.
+# config is updated with the default characters if not defined in the config.
+tokenizer, config = TTSTokenizer.init_from_config(config)
+
+# LOAD DATA SAMPLES
+# Each sample is a list of ```[text, audio_file_path, speaker_name]```
+# You can define your custom sample loader returning the list of samples.
+# Or define your custom formatter and pass it to the `load_tts_samples`.
+# Check `TTS.tts.datasets.load_tts_samples` for more details.
+train_samples, eval_samples = load_tts_samples(
+    dataset_config,
+    eval_split=True,
+    eval_split_max_size=config.eval_split_max_size,
+    eval_split_size=config.eval_split_size,
+)
+
+# init speaker manager for multi-speaker training
+# it maps speaker-id to speaker-name in the model and data-loader
+# speaker_manager = SpeakerManager()
+# speaker_manager.set_ids_from_data(train_samples + eval_samples, parse_key="speaker_name")
+# config.num_speakers = speaker_manager.num_speakers
+
+
+
+# init model
+model = Vits(config, ap, tokenizer, speaker_manager=None)
+
+# init the trainer and 🚀
+
+trainer = Trainer(
+    TrainerArgs(use_accelerate=True),
+    config,
+    output_path,
+    model=model,
+    train_samples=train_samples,
+    eval_samples=eval_samples,
+)
+trainer.fit()
+
+

trainer_0_log.txt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2d1b3ed7951f6a913b2e1bf1c3efef5eda4b3afe4b10ebcc201412aa7e6b0c92
+size 10392834