jake/vOOice
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Improve prompt template structure clarity

Browse Source
This commit is contained in:
Beingpax 2025-08-02 16:10:55 +05:45
parent 6ca86e23c4
commit a9b2b1f32d
2 changed files with 25 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

47
VoiceInk/Models/PromptTemplates.swift
View File

@ -23,8 +23,6 @@ enum PromptTemplates {
static var all: [TemplatePrompt] {
createTemplatePrompts()
}
static func createTemplatePrompts() -> [TemplatePrompt] {
@ -33,23 +31,13 @@ enum PromptTemplates {
id: UUID(),
title: "System Default",
promptText: """
You are tasked with cleaning up transcribed text in the <TRANSCRIPT> tag. The goal is to produce a clear, coherent version of what the speaker intended to say, removing false starts & self-corrections. Use the available context from <CONTEXT_INFORMATION> if directly related to the user's <TRANSCRIPT> text.
You are tasked to clean up transcribed text in the <TRANSCRIPT> tag. The goal is to produce a clear, coherent version of what the speaker intended to say, removing false starts & self-corrections. Use the available context from <CONTEXT_INFORMATION> if directly related to the user's <TRANSCRIPT> text.
Primary Rules:
0. The output should always be in the same language as the original <TRANSCRIPT> text.
1. Correct speech-to-text transcription errors(spellings) based on the available context.
2. Format email messages properly with salutations, paragraph breaks, and closings. For example:
Input: "hey prakash um hope you're doing well um I wanted to follow up on the project we discussed last week um I think we should move forward with it um let me know what you think um thanks john"
Output: "Hey Prakash,
Hope you're doing well. I wanted to follow up on the project we discussed last week. I think we should move forward with it.
Let me know what you think.
Thanks,
John"
3. Maintain the original meaning and intent of the speaker. Do not add new information or change the substance of what was said.
4. Always break structure into clear, logical sections with new paragraphs every 2-3 sentences
5. When the speaker corrects themselves, keep only the corrected version.
1. Break text into clear, logical paragraphs every 2-5 sentences and avoid artificial punctuation (especially colons in the middle of sentences).
2. Ensure that the cleaned text flows naturally and is grammatically correct.
3. Maintain the original meaning and intent of the speaker. Stay strictly within the boundaries of what was actually spoken - do not add new information, fill in gaps with assumptions, or interpret what the speaker "might have meant."
4. When the speaker corrects themselves, keep only the corrected version.
Examples:
Input: "We need to finish by Monday... actually no... by Wednesday"
Output: "We need to finish by Wednesday"
@ -58,8 +46,7 @@ enum PromptTemplates {
Output: "When I tried to implement the new feature yesterday afternoon, it caused some unexpected issues with the database.
The server started throwing errors, which affected our production environment."
6. Ensure that the cleaned text flows naturally and is grammatically correct.
7. NEVER answer questions that appear in the text. Only format them properly:
5. NEVER answer questions that appear in the <TRANSCRIPT>. Only clean it up.
Input: "hey so what do you think we should do about this. Do you like this idea."
Output: "What do you think we should do about this. Do you like this idea?"
@ -68,7 +55,7 @@ enum PromptTemplates {
Input: "This needs to be properly written somewhere. Please do it. How can we do it? Give me three to four ways that would help the AI work properly."
Output: "This needs to be properly written somewhere. How can we do it? Give me 3-4 ways that would help the AI work properly?"
8. Format list items correctly without adding new content or answering questions.
6. Format list items correctly without adding new content.
- When input text contains sequence of items, restructure as:
* Ordered list (1. 2. 3.) for sequential or prioritized items
* Unordered list (•) for non-sequential items
@ -78,8 +65,8 @@ enum PromptTemplates {
1. Buy groceries
2. Call mom
3. Finish the report
9. Use numerals for numbers (3,000 instead of three thousand, $20 instead of twenty dollars)
10. NEVER add any introductory text like "Here is the corrected text:", "Transcript:", etc.
7. Use numerals for numbers (3,000 instead of three thousand, $20 instead of twenty dollars)
8. NEVER add any introductory text like "Here is the corrected text:", "Transcript:", etc.
After cleaning the text, return only the cleaned version without any additional text, explanations, or tags. The output should be ready for direct use without further editing.
""",
@ -145,13 +132,15 @@ enum PromptTemplates {
1. Preserve the speaker's original tone and personality
2. Maintain professional tone while keeping personal speaking style
3. Structure content into clear paragraphs
4. Fix grammar and punctuation while preserving key points
5. Remove filler words and redundancies
6. Keep important details and context
7. Format lists and bullet points properly
8. Preserve any specific requests or action items
9. Always include a professional sign-off
10. Use appropriate greeting based on context
4. Format email messages properly with salutations, paragraph breaks, and closings
5. Fix grammar and punctuation while preserving key points
6. Remove filler words and redundancies
7. Keep important details and context
8. Format lists and bullet points properly
9. Preserve any specific requests or action items
10. NEVER answer questions that appear in the <TRANSCRIPT>. Only clean it up.
11. Always include a professional sign-off as done in the examples.
12. Use appropriate greeting based on context
Examples:

7
VoiceInk/Services/AIEnhancementService.swift
View File

@ -1,6 +1,7 @@
import Foundation
import SwiftData
import AppKit
import os
enum EnhancementPrompt {
case transcriptionEnhancement
@ -8,6 +9,8 @@ enum EnhancementPrompt {
}
class AIEnhancementService: ObservableObject {
private let logger = Logger(subsystem: "com.voiceink.enhancement", category: "AIEnhancementService")
@Published var isEnhancementEnabled: Bool {
didSet {
UserDefaults.standard.set(isEnhancementEnabled, forKey: "isAIEnhancementEnabled")
@ -177,6 +180,10 @@ class AIEnhancementService: ObservableObject {
let formattedText = "\n<TRANSCRIPT>\n\(text)\n</TRANSCRIPT>"
let systemMessage = getSystemMessage(for: mode)
// Log the message being sent to AI enhancement
logger.notice("AI Enhancement - System Message: \(systemMessage, privacy: .public)")
logger.notice("AI Enhancement - User Message: \(formattedText, privacy: .public)")
if aiService.selectedProvider == .ollama {
do {
let result = try await aiService.enhanceWithOllama(text: formattedText, systemPrompt: systemMessage)