DEV Community: Mateus Valentim

My React Base :)

Mateus Valentim — Sat, 22 Feb 2025 18:49:52 +0000

Hi, this is my ReactJS base. Usually, I have to build this from the ground up, but it's really tiring. Repeating everything every time is not cool. So, I created this ReactJS base for myself, and I want to share it with you 🫠.

🚀 How to Use

First, you need to clone the React base using this command:

git clone https://github.com/mattsu014/reactjs-base.git

Next, open your project folder using this command:

cd reactjs-base

Now, you need to install the node_modules using this command:

npm install

And finally, install styled-components:

npm install styled-components

📂 My React Organization

📄 Pages

The Pages folder is self-explanatory. Here, I put all the code for each page of my website.

🛠️ Components

In the Components folder, I put all the "building blocks" for my code, such as Cards, Headers, and others.

Okay, that's all! This is my simple ReactJS base. You can use your creativity to create more folders and organize your code better.

Thanks for reading 🤗.

VOSK the Offline Speech Recognition

Mateus Valentim — Fri, 31 Jan 2025 01:12:24 +0000

What is VOSK? VOSK is a powerful tool for real-time speech recognition that does not require an internet connection. Developed by Alpha Cephei, it supports multiple languages and is highly efficient. It can even run on low-performance devices, such as the Raspberry Pi.

Using VOSK

To use VOSK, you first need to download the model from this website: https://alphacephei.com/vosk/models

VOSK offers models for many languages, including Portuguese, English, Japanese, and others:

Once you download the ZIP file containing the VOSK model, you will need to unzip it

Install Dependences

First, you need to install the VOSK package and PyAudio using the following command:

pip install vosk pyaudio

PyAudio is a library that allows you to capture and reproduce audio using the PortAudio API. In this code, PyAudio is used to:

Open the microphone.
Capture audio in real time.
Collect audio data to Vosk for processing.

Let's Code!!!

Code Exemple:

import pyaudio
import json
from vosk import Model, KaldiRecognizer

model = Model("vosk-model-en-us-0.22")
recognizer = KaldiRecognizer(model, 16000)

p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True, frames_per_buffer=8192)
stream.start_stream()

print("listening...")

while True:
    data = stream.read(4096, exception_on_overflow=False)
    if recognizer.AcceptWaveform(data):
        result = json.loads(recognizer.Result())
        print("You:", result["text"])

I created this simple example code to demonstrate how VOSK works.

1. Library Imports

import pyaudio
import json
from vosk import Model, KaldiRecognizer

It's a simple import library

2. Speech recognition model loading

model = Model("vosk-model-en-us-0.22")

This loads the VOSK model for English (US).
3. Speech recognizer initialization

recognizer = KaldiRecognizer(model, 16000)

KaldiRecognizer: Starts Speech Recognition model with 16000 Hz (16 kHz) sampling rate, commum sampling rate for Speech Recognition

4. Audio Settings

p = pyaudio.PyAudio()
stream = p.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True, frames_per_buffer=8192)
stream.start_stream()

pyaudio.PyAudio(): Initializes the PyAudio object, which is used to configure and manage audio capture.
p.open(): Opens an audio stream with the following settings:
format=pyaudio.paInt16: Audio format of 16 bits.
channels=1: Mono audio (1 channel).
rate=16000: Sampling rate of 16000 Hz.
input=True: Indicates that the stream will be used for audio input (capture).
frames_per_buffer=8192: Size of the audio buffer.
stream.start_stream(): Starts the audio capture.

5. Speech Recognition Loop

print("listening...")

while True:
    data = stream.read(4096, exception_on_overflow=False)
    if recognizer.AcceptWaveform(data):
        result = json.loads(recognizer.Result())
        print("You:", result["text"])

stream.read(4096, exception_on_overflow=False):
Reads 4096 frames of audio from the stream. The exception_on_overflow=False parameter prevents the program from raising an exception in case of overflow.
recognizer.AcceptWaveform(data): Sends the audio data to the speech recognizer. If the recognizer detects a complete phrase, it returns True.
json.loads(recognizer.Result()): Converts the speech recognition result (which is in JSON format) into a Python dictionary.

Conclusion

VOSK is a powerful and efficient tool for real-time speech recognition, supporting multiple languages and running seamlessly on low-performance devices like the Raspberry Pi.
Its offline capability makes it ideal for applications where internet access is limited.
With this guide, you’ve learned how to set up VOSK, configure audio capture, and implement a basic speech recognition system.
Whether for voice-controlled apps, transcription tools, or language learning, VOSK offers a simple yet robust solution for integrating speech recognition into your projects.

Thanks for reading 🙃

Run DeepSeek R1 Locally with Ollama and Python

Mateus Valentim — Thu, 30 Jan 2025 04:36:49 +0000

In recent years, artificial intelligence and machine learning have revolutionized how we approach complex problems across various domains, from natural language processing to computer vision. In this context, tools like DeepSeek R1 have stood out for their ability to deliver efficient and high-performance solutions to challenging tasks that other models struggle with.

What is DeepSeek R1?

DeepSeek R1 is an artificial intelligence platform designed to simplify the implementation of deep learning models in real-world applications. It offers a variety of features, such as real-time data processing, custom model training, and efficiency—all wrapped in open-source code.

DeepSeek Locally and Ollama 🦙

There are many ways to use DeepSeek, one of which is through an API Key. However, here we will focus on running it locally using Ollama and Python.

First, what is Ollama? Simply put, Ollama is a tool that makes it easy to use advanced artificial intelligence models, such as those that generate text or answer questions, without requiring complex technical knowledge. It allows anyone to run and interact with these models easily, whether on a personal computer or in the cloud, making AI technology more accessible for everyday tasks or creative projects.

To install Ollama, you need to visit this website: https://ollama.com/download, select your operating system, run the installer, and complete the setup. If everything is set up correctly, execute ollama help in your terminal, and you should see this:

Large language model runner

Usage:
  ollama [flags]
  ollama [command]

Available Commands:
  serve       Start ollama
  create      Create a model from a Modelfile
  show        Show information for a model
  run         Run a model
  stop        Stop a running model
  pull        Pull a model from a registry
  push        Push a model to a registry
  list        List models
  ps          List running models
  cp          Copy a model
  rm          Remove a model
  help        Help about any command

Flags:
  -h, --help      help for ollama
  -v, --version   Show version information

Use "ollama [command] --help" for more information about a command.

The next step is to download the model. On the Ollama website's search page https://ollama.com/search, you can find many models to use.

In this post, we will use DeepSeek R1 models https://ollama.com/library/deepseek-r1:

DeepSeek-R1-Distill-Qwen-1.5B
ollama run deepseek-r1:1.5b

DeepSeek-R1-Distill-Qwen-7B
ollama run deepseek-r1:7b

DeepSeek-R1-Distill-Llama-8B
ollama run deepseek-r1:8b

DeepSeek-R1-Distill-Qwen-14B
ollama run deepseek-r1:14b

DeepSeek-R1-Distill-Qwen-32B
ollama run deepseek-r1:32b

DeepSeek-R1-Distill-Llama-70B
ollama run deepseek-r1:70b

Each model differs in terms of its parameters. Essentially, the parameters of an AI model are like "internal settings" that it adjusts to learn from data and make predictions. The more parameters a model has, the more complex and capable it can be, but it also requires more processing power.

Note: Computational power is very important when choosing a model. The higher the number of parameters, the more computational power your machine will need.

After downloading the model, your chat will show this:

>>> Send a message (/? for help)

Now you have a local AI on your computer.

Ollama and Python 🐍

To connect Python and Ollama with DeepSeek, first, you need to create your Python file and install the Ollama package using this command:

pip install ollama

I created the following code as an example:

import ollama

model = "deepseek-r1:7b"

while True:
    Myprompt = input("You: ")
    answer = ollama.generate(model=model, prompt=Myprompt)
    print("DeepSeek:", answer["response"])

Import the package using import ollama.
Define your model using model = "deepseek-r1:7b".
Start a loop with while True:.
Capture user input using prompt = input("You: ").
Generate a response from the model using answer = ollama.generate(model=model, prompt=prompt).
Print the response using print("DeepSeek:", answer["response"]).

Conclusion

DeepSeek R1, combined with Ollama, offers a powerful yet accessible way to run advanced AI models locally on your machine. This setup allows you to experiment with AI technology without the need for extensive infrastructure or cloud resources.

By following the steps outlined in this article, you can easily install Ollama, download DeepSeek R1 models, and integrate them with Python for seamless interaction. This approach not only simplifies the deployment process but also opens up opportunities for customization and experimentation in your AI projects.

Thanks for reading 🙃