[Day 212] Final Glaswegian TTS model

 Hello :)
Today is Day 212!


A quick summary of today:
  • creating a simple Glaswegian assistant app


After finetuning whisper to get the final version of the Glaswegian ASR model, the next task was to do a final fine-tuning on the T5 speech model to get a final version of the Glaswegian TTS model. Well I did that today. 

Here is a link to the model on HuggingFace. And its training results:

Now that we have the final 2 hour dataset, I was hoping for better results. Before, the generated audio (while with a little accent) sounded robotic. First thing I had to do was fix the HuggingFace space where the previous version of the glaswegian TTS was running. The issue was related to voice embeddings, and after a quick fix ~

It was up again, and I loaded the latest glaswegian_tts model. Well now, it *does* sound better. There are cases where it is robotic, but there is definitely improvement compared to the previous version. That previous version was trained or around 30 mins of audio, compared to now 2 hours.


Next - create a full assistant app

Gradio and HF spaces make it very easy -> here.
Audio input -> transcribed using glaswegian_asr -> send to gpt2 -> answer from gpt2 is turned to speech and returned to the user

At the start I used gpt2, but as its not that good, I switched to using gpt3.5-turbo.


That is all for today!

See you tomorrow :)

Popular posts from this blog

[Day 198] Transactions Data Streaming Pipeline Porject [v1 completed]

Image
 Hello :) Today is Day 198! A quick summary of today: data streaming pipeline project [v1 done] Here is a link to the project's repo. Well ... I did not know I can do it in a day (~14 hours) after yesterday's issues but here we are. Turns out in order to insert the full (~70 variables with nested/list structure), I need the proper pyspark schema. And yesterday I did not have that and that is why when I was reading data in the kafka producer I was getting NULL in the columns - my schema was wrong. Well today I not only fixed the schema for the 4 variables I had yesterday, but included *all* the variables that come from the Stripe API ~ 70 (for completeness).  When I run docker-compose, the data streams and is input into the postgres db (and is still running). Unfortunately, the free Stripe API for creating realistic transactions has a limit of 25, so every 3 seconds, 25 new transactions are sent to the db. It has been running half the day (since I got that set up) and as I am w...
Read more

[미리 공부] 기초 통계 복습 (Day 1는 1월2일)

Image
모험을 시작하기 전에 기초 지식을 복습하고자 했다. 오늘 SPSS 대안 프로그램을 찾아보려고 해서 JASP에 대해 알게 되었다. 유용한 프로그램인 것 같아서 선회귀와 기술통계를 내려고 했는데 재미있었다. 그런데 JASP에 대해 더 알기 전에 '기초 통계 지식을 좀 복습을 하고자 하면 좋을 것 같아'란 생각을 들었다. 다행히, Coursera에서 Stanford University의 Guenther Walther 교수님께서 진행된  Introduction to Statistics  무료 강좌가 있다.  좀 부족한 부분은 다양한 검정통계 하는 거고 (F test, t-test, chi-square 등) 이제 JASP 아니면 다른 통계 프로그램 사용하게 되어도 이런 부분을 좀 더 자세히 집중하여 공부하면 된다.  특히 homoscedasticity 및 heteroscedasticity 개념을 기억에 남았다. Homoscedasticity (선): Definition: In a homoscedastic dataset, the variance of the errors (residuals) is constant across all levels of the independent variable(s). In simpler terms, the spread of the residuals is the same throughout the range of predictor values. Heteroscedasticity (악): Definition: Heteroscedasticity occurs when the variance of the errors is not constant across all levels of the independent variable(s). In other words, the spread of residuals changes as the values of the independe...
Read more

[Day 61] Stanford CS224N (NLP with DL): Machine translation, seq2seq + a side CDCGAN mini project

Image
 Hello :) Today is Day 61! A quick summary of today: Covered Lecture 7 : machine translation, seq2seq, attention from Stanford CS224N Tried to make a conditional DCGAN to generate MNIST numbers ( colab ) ( kaggle ) I will first cover the GAN story (then will share my notes from the lecture) So... while watching and taking notes today, I started thinking, what if I can use my notes as data to a model and afterwards, when I want, I can give it raw string text and it will output text in the format of my notes (with my handwriting). Well I started looking around and actually the first model architecture that came to my mind was the GAN (specifically conditional GAN) - I remembered there was a GAN architecture that alongside the pictures, we can give it the labels, and then on-demand generate. In retrospect, there are of course others, but I decided to go with GAN.  For maybe 2 hours I busted my head trying to make a simple model with the EMNIST dataset (english characters), and I ...
Read more