CaraDuf

Why is bigvgan better than Avovodo ?

2

Hi, I finetune Toucan Meta model on 1k on a reduced dataset to understand the difference between Avocodo and BigVGan. Here are the spectrograms :  Apart from the 12kHz...

New ToucanTTS model gives far worse results after finetuning

13

Hi, I tried to finetune the new Meta model on a 87 sample dataset in French that I used several times already but now the results is very bad. I...

How to read scorer output ?

5

Hi, I want to check what the scorer has to say about my dataset and why it is keeping only 77 samples out of 98 (which all sound ok to...

Use Pytorch Lightning to speed up training ?

3

Hi, In a previous [answer](https://github.com/DigitalPhonetics/IMS-Toucan/issues/109#issuecomment-1475011361) you wrote that you were looking for ways to improve training speed even though you were already satisfied with Toucan's training performance. Have you ever...

Why does utterance cloner need to know the reference transcription text ?

3

Hi, I tried the run_utterance_cloner and noticed very bad results when the transcription text does not match the reference audio. In another project I tried (Coqui) that also does voice...

Number of fine tuning steps recommended when fine tuning to avoid overfitting

3

Hi, Given a target speaker dataset what is roughly the number of fine tuning steps that should be undergone ? [NeMo](https://github.com/NVIDIA/NeMo/blob/main/tutorials/tts/FastPitch_Finetuning.ipynb) "recommends 1000 steps per minute of audio for fastpitch...

Read text from a specific speaker seen during training/finetuning ?

2

Hi, I merged all my single speaker datasets into a bigger one and finetuned the Meta model on it. Now when inferencing the output sounds like a mixture of all...

Where can I intercept the Mel Spectrogram to save it as .npy ?

2

Hi, Out of curiosity, I want to test BigVGan. On their [page](https://github.com/NVIDIA/BigVGAN) they say that it accepts `.npy` as input. I browsed the code but could not find where the...

How long in seconds should the speaker reference audio be ?

2

Hi, When using `read_texts` function how long should the `speaker_reference` be and how should it be to give best results ? By "how long" I mean its duration in seconds...

Running weight averaging during training seems to hang it

2

Hi, I am running the training in one remote terminal and I am doing inference of the current model on another one. Sometimes I test the current model and to...