Need help in training for telugu language

[shashidhar2609]

Hi,

I have been trained telugu text to speech using keithio tacotron previously by converting telugu text to transliterated english text ,it was working good but for longer sentences it was not performing well.i found your model aligns good for even longer sentences.so i was trying to train using ljspeech-ddc .i have 30 hours of single speaker data and corresponding transliterated text. even though model trained for 17000 steps the output of synthesis.py is null. could you please help by letting me know where i am going wrong.i have set "use_phonemes": false.

command :python synthesize.py --text "anndaatku anni veelllloo andubaattuloo unddeenduku raassttr prbhutvn" --config_path /root/tacotron/LJSpeech/ljspeech-ddc-July-14-2021_08+44PM-e9e0784/config.json --model_path /root/tacotron/LJSpeech/ljspeech-ddc-July-14-2021_08+44PM-e9e0784/checkpoint_14000.pth.tar --out_path ./predicted
2021-07-15 10:10:16.604787: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.10.1

Setting up Audio Processor...
| > sample_rate:22050
| > resample:False
| > num_mels:80
| > min_level_db:-100
| > frame_shift_ms:None
| > frame_length_ms:None
| > ref_level_db:20
| > fft_size:1024
| > power:1.5
| > preemphasis:0.0
| > griffin_lim_iters:60
| > signal_norm:True
| > symmetric_norm:True
| > mel_fmin:50.0
| > mel_fmax:8000.0
| > spec_gain:20.0
| > stft_pad_mode:reflect
| > max_norm:4.0
| > clip_norm:True
| > do_trim_silence:True
| > trim_db:60
| > do_sound_norm:False
| > stats_path:None
| > hop_length:256
| > win_length:1024
Using model: Tacotron2
Text: anndaatku anni veelllloo andubaattuloo unddeenduku raassttr prbhutvn
Text splitted to sentences.
['anndaatku anni veelllloo andubaattuloo unddeenduku raassttr prbhutvn']
Processing time: 0.25736236572265625
Real-time factor: 0.4612191290787199
Saving output to ./predicted/anndaatku_anni_veell.wav

the output anndaatku_anni_veell.wav has nothing in it.

training paramteres:
12|tts_train | --> STEP: 654/850 -- GLOBAL_STEP: 17675
12|tts_train | | > decoder_loss: 0.49871 (0.48944)
12|tts_train | | > postnet_loss: 0.55522 (0.55548)
12|tts_train | | > stopnet_loss: 2.01851 (2.04752)
12|tts_train | | > decoder_coarse_loss: 0.73749 (0.70555)
12|tts_train | | > decoder_ddc_loss: 0.01430 (0.02146)
12|tts_train | | > ga_loss: 0.00012 (0.00019)
12|tts_train | | > decoder_diff_spec_loss: 0.24318 (0.24266)
12|tts_train | | > postnet_diff_spec_loss: 0.32557 (0.32943)
12|tts_train | | > decoder_ssim_loss: 0.56306 (0.50855)
12|tts_train | | > postnet_ssim_loss: 0.58082 (0.52771)
12|tts_train | | > loss: 1.19283 (1.15014)
12|tts_train | | > align_error: 0.52511 (0.51386)
12|tts_train | | > max_spec_length: 601.0
12|tts_train | | > max_text_length: 112.0
12|tts_train | | > step_time: 3.2762
12|tts_train | | > loader_time: 0.01
12|tts_train | | > current_lr: 0.0001

Thanks

[rajashekhargithubb]

Hello did you found anything?

[RAVINDRA8008]

bro got any solution??? i need i am doing a project so?