vakyansh-tts

vakyansh-tts copied to clipboard

Model can't speak decimal numbers in Hindi Model

1

When trying to to give a float number in text, model crashes. It shows this error ` assert digit in all_num["en"], "Give proper input" AssertionError: Give proper input `

mediocremonke

Hi, Great work. I am a bit confused, can we done fine tuning with custom dataset of single speaker? If so, how, please guide.

naveed81

default_lineup.json is not available File "/home/vivek/ml/tushar/indic_tts/vakyansh-tts/utils/inference/transliterate.py", line 720, in __init__ lineup = json.load(open(os.path.join(F_DIR, config_path), encoding="utf-8"))

tushar7770

error in tts generation for number '1998' in Tamil

4

error in tts generation for number '1998' in Tamil; it should speak "ஓர் ஆயிரத்து தொள்ளாயிரத்து தொன்னூற்றெட்டு" as in https://tamilpesu.us/en/number/1998/?type=IN current implementation in vakyansh-tts speaks incorrectly

arcturusannamalai

gpu usage during inference

1

Hi is it compulsory to use GPU during inference while using your models.

kavya2425

Can you please guide on what parameters need to be changed to start multispeaker training. I have dataset of 4 speakers. I have transformed the data into required format using...

14besesraza

Is there a way to add voice cloning while synthesizing speech, something similar to [SV2TTS](https://arxiv.org/abs/1806.04558) - if I have a trained speaker encoder, is it possible to do real time...

Sujit27

When try to run "pip install -r requirements.txt" in a virtual environment in WSL, I get this error: Failed to build mosestokenizer ffmpy flask-cachebuster docopt toolwrapper uctools Could you help...

PS-AI

Hi, When I was trying to finetune using the pre-trained checkpoint of English Glow model it is causing the grad to become `inf/ Nan` inspite of the defined gradient clipping...

Aalisha