Playground Documentation

Welcome to the SipPulse AI Playground! Here you can experiment with different AI models and adjust various settings to see how they affect the results. This guide will help you navigate and use the Playground effectively.

Overview

The Playground allows you to:

• Test AI Models: Select and execute different AI models.
• Adjust Parameters: Modify settings such as temperature, maximum tokens, Top P, frequency penalty, and presence penalty.
• Manage Messages: Add system and user messages to interact with the model.
• Convert Speech to Text: Use the "Speech to Text" tab to transcribe audio files.
• Speech Analytics: Anonimyze, Summarize, Sentiment Analysis, Topic Detection, and Custom Analysis
• Convert Text to Speech: Use the "Text to Speech" tab to generate audio from text.
• Anonymize Text: Anonymize (redact) text to remove confidential information
• Check Pricing: Consult the "Pricing" tab for details on the costs associated with using the models.
• View Code: See how to access the APIs in Curl, Python, and JavaScript.

Navigating the Playground

Model Selector

In the Model field, you can choose from various available models. The current models include:

• qwen-max
• qwen-plus
• qwen-turbo
• gpt-3.5-turbo
• gpt-4-turbo
• gpt-4o
• gpt-4o-mini
• claude-3.7-haiku
• claude-3.7-sonnet
• deepseek-v3
• llama-3.3-70b
• llama-3.1-405b (Alpha)
• pulse-precision (zero hallucination low cost speech-to-text transcription)
• pulse-speed(serverless zero hallucination high speed STT transcription)
• whisper-1 (open AI whisper-1)
• whisper-chat (ultra-fast chat oriented STT)
• azure-tts (for text-to-speech conversion)

Text Generation Options

Temperature

• Description: Controls the randomness of the model's responses.
• How to Use: A lower value (e.g., 0.2) makes the responses more deterministic and repetitive, while a higher value (e.g., 0.8) makes the responses more varied and creative.
• Setting: Use the slider to adjust the temperature.

Max Tokens

• Description: Defines the maximum number of tokens that the model's response can have.
• How to Use: Increase for longer responses or decrease for shorter responses.
• Setting: Use the slider to adjust the maximum number of tokens.

Top P

• Description: Also known as nucleus sampling, this parameter controls the diversity of responses by considering the cumulative probability of token options.
• How to Use: A value of 1 considers all tokens, while a lower value (e.g., 0.9) considers only the tokens that make up the cumulative probability up to that point.
• Setting: Use the slider to adjust the Top P.

Frequency Penalty

• Description: Penalizes new tokens based on their frequency so far.
• How to Use: Increasing this value can reduce the repetition of words.
• Setting: Use the slider to adjust the frequency penalty.

Presence Penalty

• Description: Penalizes new tokens based on whether they appear in the text so far.
• How to Use: Increasing this value can reduce the repetition of ideas.
• Setting: Use the slider to adjust the presence penalty.

Message Sections

System Message

• Description: Defines the context or role that the model should assume.
• How to Use: Enter a prompt such as "You are a helpful AI assistant."
• Field: System

User Message

• Description: Enter the message that the user would send to interact with the model.
• How to Use: Write the question or command you want the model to respond to.
• Field: User

To add a message, enter the text in the appropriate field and click Add message.

Execute Generation

After configuring all parameters and messages, click the Run button to execute the model and generate the response.

Speech to Text Conversion

The Speech to Text tab allows you to transcribe audio files to text. Here's how to use it:

Model Selector

In the Model field, select the available transcription model:

• whisper-1

Transcription Options

Output Format

• Text: Transcribes the audio to text.

Language

• Description: Selects the language of the audio for transcription.
• How to Use: Choose the language of the audio to be transcribed from the dropdown menu.

Prompt

• Description: Enter a prompt to provide additional context for the transcription, if necessary.
• Field: Prompt

Upload Audio File

• Description: Drag and drop an audio file into the designated area to upload it.
• Supported Formats: Ensure the audio file is in a supported format.

Execute Transcription

After configuring the options and uploading the audio file, click the Transcribe button to start the transcription.

Text to Speech Conversion

The Text to Speech tab allows you to convert text to audio. Here's how to use it:

Model Selector

In the Model field, select the available conversion model:

• azure-tts

Conversion Options

Language

• Description: Select the language of the generated speech.
• How to Use: Choose the desired language from the dropdown menu.
• Options: Various languages are available, such as English (US), Spanish, etc.

Voice

• Description: Choose the voice that the model should use for the conversion.
• How to Use: Select the desired voice from the dropdown menu.
• Options: Available voices vary based on the selected language.

Output Format

• Description: Choose the audio format for the output.
• How to Use: Select the desired format from the dropdown menu.
• Options: Formats such as MP3, WAV, etc.

Text

• Description: Enter the text you want to convert to speech.
• How to Use: Type or paste the text into the text field.

Execute Conversion

After configuring the options and entering the text, click the Run button to start the text-to-speech conversion.

View Code

The View Code tab provides examples of how to integrate the current settings and prompt into your application using Curl, Python, and JavaScript.

Curl Example

curl -X 'POST' \
  'https://api.sippulse.ai/v1/tts/generate' \
  -H 'Content-Type: application/json' \
  -H 'api-key: $SIPPULSE_API_KEY' \
  -d '{
  "text": "This is a text to speech generation",
  "voice": "en-US-AvaMultilingualNeural",
  "output_format": 5,
  "model": "azure-tts"
}'

Python Example

pip install requests

import requests

url = "https://api.sippulse.ai/v1/tts/generate"
headers = {
    "Content-Type": "application/json",
    "api-key": "YOUR_SIPPULSE_API_KEY"
}
data = {
    "text": "This is a text to speech generation",
    "voice": "en-US-AvaMultilingualNeural",
    "output_format": 5,
    "model": "azure-tts"
}

response = requests.post(url, headers=headers, json=data)
audio_content = response.content

with open("output.mp3", "wb") as audio_file:
    audio_file.write(audio_content)

Javascript Example

npm install node-fetch

const fetch = require('node-fetch');

const url = 'https://api.sippulse.ai/v1/tts/generate';
const apiKey = 'YOUR_SIPPULSE_API_KEY';

const data = {
  text: "This is a text to speech generation",
  voice: "en-US-AvaMultilingualNeural",
  output_format: 5,
  model: "azure-tts"
};

const options = {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'api-key': apiKey
  },
  body: JSON.stringify(data)
};

fetch(url, options)
  .then(response => response.arrayBuffer())
  .then(buffer => {
    const fs = require('fs');
    fs.writeFile('output.mp3', Buffer.from(buffer), () => {
      console.log('Audio file saved as output.mp3');
    });
  })
  .catch(err => console.error('Error:', err));

---

We hope you enjoy using the SipPulse AI Playground to explore the capabilities of our AI models. If you have any questions, please refer to our Complete Documentation or contact support.

Happy experimenting!