search

coqui-ai / TTS

🐸💬 - a deep learning toolkit for Text-to-Speech, battle-tested in research and production
http://coqui.ai
Mozilla Public License 2.0
35.87k stars 4.39k forks source link

[Bug] xtts OrderedVocab problem #3114

Closed jazza420 closed 1 year ago

jazza420 commented 1 year ago

Describe the bug

TRAINING (2023-10-28 18:37:37) The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted ! The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !

To Reproduce

training xtts with standard recipe

Expected behavior

No response

Logs

python finetunextts.py
>> DVAE weights restored from: C:\Users\someone\Desktop\xtts/run\training\XTTS_v1.1_original_model_files/dvae.pth
 | > Found 489 files in C:\Users\someone\Desktop\xtts
fatal: not a git repository (or any of the parent directories): .git
fatal: not a git repository (or any of the parent directories): .git
 > Training Environment:
 | > Current device: 0
 | > Num. of GPUs: 1
 | > Num. of CPUs: 12
 | > Num. of Torch Threads: 1
 | > Torch seed: 1
 | > Torch CUDNN: True
 | > Torch CUDNN deterministic: False
 | > Torch CUDNN benchmark: False
 > Start Tensorboard: tensorboard --logdir=C:\Users\someone\Desktop\xtts/run\training\GPT_XTTS_LJSpeech_FT-October-27-2023_10+56PM-0000000

 > Model has 543985103 parameters

 > EPOCH: 0/1000
 --> C:\Users\someone\Desktop\xtts/run\training\GPT_XTTS_LJSpeech_FT-October-27-2023_10+56PM-0000000
 > Filtering invalid eval samples!!
 > Total eval samples after filtering: 4

 > EVALUATION

 | > Synthesizing test sentences.

  --> EVAL PERFORMANCE
     | > avg_loader_time: 0.01900 (+0.00000)
     | > avg_loss_text_ce: 0.04067 (+0.00000)
     | > avg_loss_mel_ce: 4.33739 (+0.00000)
     | > avg_loss: 4.37806 (+0.00000)

 > BEST MODEL : C:\Users\someone\Desktop\xtts/run\training\GPT_XTTS_LJSpeech_FT-October-27-2023_10+56PM-0000000\best_model_0.pth

 > EPOCH: 1/1000
 --> C:\Users\someone\Desktop\xtts/run\training\GPT_XTTS_LJSpeech_FT-October-27-2023_10+56PM-0000000
 > Sampling by language: dict_keys(['en'])

 > TRAINING (2023-10-27 22:57:08)
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5024, your vocabulary could be corrupted !
The OrderedVocab you are attempting to save contains a hole for index 5025, your vocabulary could be corrupted !

   --> STEP: 0/243 -- GLOBAL_STEP: 0
     | > loss_text_ce: 0.04536  (0.04536)
     | > loss_mel_ce: 4.79820  (4.79820)
     | > loss: 4.84356  (4.84356)
     | > current_lr: 0.00001
     | > step_time: 0.88430  (0.88431)
     | > loader_time: 70.14840  (70.14841)

   --> STEP: 50/243 -- GLOBAL_STEP: 50
     | > loss_text_ce: 0.04994  (0.04525)
     | > loss_mel_ce: 5.39171  (4.74854)
     | > loss: 5.44165  (4.79379)
     | > current_lr: 0.00001
     | > step_time: 0.66870  (1.54556)
     | > loader_time: 0.01600  (0.01624)

   --> STEP: 100/243 -- GLOBAL_STEP: 100
     | > loss_text_ce: 0.04045  (0.04345)
     | > loss_mel_ce: 3.84910  (4.67366)
     | > loss: 3.88955  (4.71711)
     | > current_lr: 0.00001
     | > step_time: 1.74700  (1.66512)
     | > loader_time: 0.01520  (0.01434)

   --> STEP: 150/243 -- GLOBAL_STEP: 150
     | > loss_text_ce: 0.05477  (0.04379)
     | > loss_mel_ce: 5.39814  (4.72587)
     | > loss: 5.45292  (4.76966)
     | > current_lr: 0.00001
     | > step_time: 2.80970  (1.85835)
     | > loader_time: 0.01400  (0.01352)

   --> STEP: 200/243 -- GLOBAL_STEP: 200
     | > loss_text_ce: 0.03867  (0.04367)
     | > loss_mel_ce: 4.21473  (4.71702)
     | > loss: 4.25340  (4.76068)
     | > current_lr: 0.00001
     | > step_time: 3.30200  (2.20536)
     | > loader_time: 0.00500  (0.01207)

 > Filtering invalid eval samples!!
 > Total eval samples after filtering: 4

 > EVALUATION

 | > Synthesizing test sentences.

  --> EVAL PERFORMANCE
     | > avg_loader_time: 0.01202 (-0.00698)
     | > avg_loss_text_ce: 0.03961 (-0.00106)
     | > avg_loss_mel_ce: 4.15599 (-0.18140)
     | > avg_loss: 4.19560 (-0.18246)

 > BEST MODEL : C:\Users\someone\Desktop\xtts/run\training\GPT_XTTS_LJSpeech_FT-October-27-2023_10+56PM-0000000\best_model_243.pth

Environment

{
    "CUDA": {
        "GPU": [
            "NVIDIA GeForce RTX 3060"
        ],
        "available": true,
        "version": "11.7"
    },
    "Packages": {
        "PyTorch_debug": false,
        "PyTorch_version": "2.0.1",
        "TTS": "0.19.0",
        "numpy": "1.22.0"
    },
    "System": {
        "OS": "Windows",
        "architecture": [
            "64bit",
            "WindowsPE"
        ],
        "processor": "AMD64 Family 25 Model 80 Stepping 0, AuthenticAMD",
        "python": "3.9.13",
        "version": "10.0.22621"
    }
}

Additional context

No response

erogol commented 1 year ago

The vocab file has missing items. It would not affect the model's performance. You can ignore the warning.