AGENT SKILL

speak-core-workflow-b

Execute Speak secondary workflow: Pronunciation Training with detailed phoneme analysis. Use when implementing pronunciation drills, speech scoring, or targeted pronunciation improvement features. Trigger with phrases like "speak pronunciation training", "speak speech scoring", "secondary speak workflow".

License 许可
MIT
最近更新
2026-11-15
Stars
★ 2.0k
Forks
233

01 前言 Front Matter

name
speak-core-workflow-b
author
Jeremy Longshore <jeremy@intentsolutions.io>
license
MIT
version
1.0.0
description
Execute Speak secondary workflow: Pronunciation Training with detailed phoneme analysis. Use when implementing pronunciation drills, speech scoring, or targeted pronunciation improvement features. Trigger with phrases like "speak pronunciation training", "speak speech scoring", "secondary speak workflow".
allowed-tools
Read, Write, Edit, Bash(npm:*), Grep

03 SKILL.md 正文

Speak Core Workflow B: Pronunciation Training

Overview

Secondary workflow for Speak: Detailed pronunciation training with phoneme-level analysis and targeted practice.

Prerequisites

  • Completed speak-install-auth setup
  • Familiarity with speak-core-workflow-a
  • Valid API credentials configured
  • High-quality audio input capabilities

Instructions

Step 1: Initialize Pronunciation Session

// src/workflows/pronunciation-training.ts
import {
  SpeakClient,
  PronunciationTrainer,
  PhonemeAnalyzer,
} from '@speak/language-sdk';

interface PronunciationConfig {
  targetLanguage: string;
  difficulty: 'beginner' | 'intermediate' | 'advanced';
  focusPhonemes?: string[]; // Specific sounds to practice
  category?: 'vowels' | 'consonants' | 'tones' | 'all';
}

async function initializePronunciationTraining(
  client: SpeakClient,
  config: PronunciationConfig
): Promise<PronunciationTrainer> {
  const trainer = new PronunciationTrainer(client, {
    language: config.targetLanguage,
    difficulty: config.difficulty,
    adaptiveMode: true, // Adjusts based on user performance
  });

  // Pre-load phoneme models for target language
  await trainer.initialize();

  // Set focus areas if specified
  if (config.focusPhonemes) {
    trainer.setFocusPhonemes(config.focusPhonemes);
  }

  return trainer;
}

Step 2: Implement Drill Session

interface DrillItem {
  id: string;
  text: string;
  romanization?: string; // For non-Latin scripts
  translation: string;
  audioUrl: string;
  targetPhonemes: string[];
  difficulty: number;
}

interface DrillResult {
  item: DrillItem;
  userAudio: ArrayBuffer;
  scores: PronunciationScores;
  phonemeDetails: PhonemeResult[];
  feedback: string[];
}

async function runPronunciationDrill(
  trainer: PronunciationTrainer,
  drillCount: number = 10
): Promise<DrillSession> {
  const session = await trainer.startDrillSession({
    itemCount: drillCount,
    repeatOnMistake: true,
    minScore: 70, // Repeat until 70% or better
  });

  const results: DrillResult[] = [];

  while (!session.isComplete) {
    // Get next drill item
    const item = await session.getNextItem();

    console.log('\n--- Pronunciation Drill ---');
    console.log(`Say: "${item.text}"`);
    if (item.romanization) {
      console.log(`Romanization: ${item.romanization}`);
    }
    console.log(`Meaning: ${item.translation}`);
    console.log('Listen to native pronunciation...');

    // Play native audio for reference
    await playAudio(item.audioUrl);

    // Record user attempt
    const userAudio = await recordUserAudio();

    // Analyze pronunciation
    const result = await session.submitAttempt({
      itemId: item.id,
      audioData: userAudio,
    });

    results.push({
      item,
      userAudio,
      scores: result.scores,
      phonemeDetails: result.phonemeDetails,
      feedback: result.feedback,
    });

    // Display detailed feedback
    displayPronunciationFeedback(result);

    // Check if need to repeat
    if (result.scores.overall < 70 && session.shouldRepeat) {
      console.log('\nLet\'s try that one again...');
    }
  }

  return session.getSummary();
}

Step 3: Phoneme-Level Analysis

interface PhonemeResult {
  phoneme: string;
  expected: string;
  actual: string;
  score: number;
  issues: PhonemeIssue[];
  visualGuide?: string; // Mouth position diagram URL
}

interface PhonemeIssue {
  type: 'substitution' | 'omission' | 'addition' | 'distortion';
  description: string;
  tip: string;
}

function displayPronunciationFeedback(result: DrillResult) {
  console.log(`\n? Pronunciation Score: ${result.scores.overall}/100`);
  console.log(`   Accuracy: ${result.scores.accuracy}/100`);
  console.log(`   Fluency: ${result.scores.fluency}/100`);
  console.log(`   Intonation: ${result.scores.intonation}/100`);

  // Show problem phonemes
  const problemPhonemes = result.phonemeDetails.filter(p => p.score < 70);
  if (problemPhonemes.length > 0) {
    console.log('\n? Phonemes to improve:');
    for (const p of problemPhonemes) {
      console.log(`   [${p.phoneme}] Score: ${p.score}/100`);
      for (const issue of p.issues) {
        console.log(`      ⚠️ ${issue.type}: ${issue.description}`);
        console.log(`      ? Tip: ${issue.tip}`);
      }
    }
  }

  // Show overall feedback
  if (result.feedback.length > 0) {
    console.log('\n? Feedback:');
    result.feedback.forEach(f => console.log(`   • ${f}`));
  }
}

Step 4: Adaptive Practice Generation

interface WeaknessAnalysis {
  phoneme: string;
  averageScore: number;
  attemptCount: number;
  trend: 'improving' | 'stable' | 'declining';
  suggestedDrills: DrillItem[];
}

async function generateAdaptivePractice(
  trainer: PronunciationTrainer,
  userHistory: DrillResult[]
): Promise<AdaptivePracticeSession> {
  // Analyze user's weaknesses
  const weaknesses = analyzeWeaknesses(userHistory);

  // Generate targeted practice
  const adaptiveSession = await trainer.createAdaptiveSession({
    targetWeaknesses: weaknesses.map(w => w.phoneme),
    intensity: 'focused', // or 'mixed' for variety
    maxDuration: 15 * 60 * 1000, // 15 minutes
  });

  console.log('\n? Adaptive Practice Plan:');
  console.log(`Focus areas: ${weaknesses.map(w => w.phoneme).join(', ')}`);

  return adaptiveSession;
}

function analyzeWeaknesses(results: DrillResult[]): WeaknessAnalysis[] {
  const phonemeStats = new Map<string, number[]>();

  // Collect scores by phoneme
  for (const result of results) {
    for (const p of result.phonemeDetails) {
      if (!phonemeStats.has(p.phoneme)) {
        phonemeStats.set(p.phoneme, []);
      }
      phonemeStats.get(p.phoneme)!.push(p.score);
    }
  }

  // Identify weaknesses (average < 75)
  const weaknesses: WeaknessAnalysis[] = [];
  for (const [phoneme, scores] of phonemeStats) {
    const avg = scores.reduce((a, b) => a + b, 0) / scores.length;
    if (avg < 75) {
      weaknesses.push({
        phoneme,
        averageScore: avg,
        attemptCount: scores.length,
        trend: calculateTrend(scores),
        suggestedDrills: [], // Populated by trainer
      });
    }
  }

  return weaknesses.sort((a, b) => a.averageScore - b.averageScore);
}

Complete Workflow Example

async function pronunciationTrainingWorkflow() {
  const client = getSpeakClient();

  // Configure for Korean pronunciation (challenging for English speakers)
  const config: PronunciationConfig = {
    targetLanguage: 'ko',
    difficulty: 'intermediate',
    focusPhonemes: ['ㄱ', 'ㅋ', 'ㄲ'], // Korean aspirated/tense consonants
    category: 'consonants',
  };

  console.log('Starting pronunciation training...');
  console.log(`Language: ${config.targetLanguage}`);
  console.log(`Focus: ${config.focusPhonemes?.join(', ') || 'General'}`);

  // Initialize trainer
  const trainer = await initializePronunciationTraining(client, config);

  // Run initial assessment
  console.log('\n? Initial Assessment...');
  const assessment = await trainer.runAssessment();
  console.log(`Baseline score: ${assessment.overallScore}/100`);

  // Run pronunciation drills
  const drillResults = await runPronunciationDrill(trainer, 10);

  // Generate adaptive practice based on results
  const adaptiveSession = await generateAdaptivePractice(
    trainer,
    drillResults.results
  );

  // Run adaptive practice
  await adaptiveSession.run();

  // Final summary
  const summary = await trainer.getSessionSummary();
  console.log('\n========== Training Complete ==========');
  console.log(`Items practiced: ${summary.totalItems}`);
  console.log(`Average score: ${summary.averageScore}/100`);
  console.log(`Improvement: +${summary.improvement}%`);

  return summary;
}

Workflow Comparison

Aspect Workflow A (Conversation) Workflow B (Pronunciation)
Primary Focus Communication Accuracy
Feedback Type Holistic Phoneme-level
Session Style Free-form dialogue Structured drills
Pacing User-driven System-driven
Best For Fluency building Accent reduction

Output

  • Detailed phoneme-level scores
  • Visual pronunciation guides
  • Adaptive practice recommendations
  • Progress tracking over time
  • Weakness identification

Error Handling

Error Cause Solution
Audio Too Short Brief recording Minimum 0.5s audio
Background Noise Poor recording conditions Prompt for quieter environment
Phoneme Not Detected Unclear speech Slow down and articulate
Model Loading Failed Network issue Retry with fallback

Resources

Next Steps

For common errors, see speak-common-errors.

04 同仓库更多 Skills

Dicklesworthstone
Dicklesworthstone/pi_agent_rust

theme-factory

Toolkit for styling artifacts with a theme. These artifacts can be slides, docs, reportings, HTML landing pages, etc. There are 10 pre-set themes with colors/fonts that you can apply to any artifact that has been creating, or can generate a new theme on-the-fly.

视觉设计
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

web-artifacts-builder

Suite of tools for creating elaborate, multi-component claude.ai HTML artifacts using modern frontend web technologies (React, Tailwind CSS, shadcn/ui). Use for complex artifacts requiring state management, routing, or shadcn/ui components - not for simple single-file HTML/JSX artifacts.

前端开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

webapp-testing

Toolkit for interacting with and testing local web applications using Playwright. Supports verifying frontend functionality, debugging UI behavior, capturing browser screenshots, and viewing browser logs.

测试与质量前端开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

neuropixels-analysis

Neuropixels neural recording analysis. Load SpikeGLX/OpenEphys data, preprocess, motion correction, Kilosort4 spike sorting, quality metrics, Allen/IBL curation, AI-assisted visual analysis, for Neuropixels 1.0/2.0 extracellular electrophysiology. Use when working with neural recordings, spike sorting, extracellular electrophysiology, or when the user mentions Neuropixels, SpikeGLX, Open Ephys, Kilosort, quality metrics, or unit curation.

★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

skill-creator

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Skill与智能体开发
★ 2.0k
Dicklesworthstone
Dicklesworthstone/pi_agent_rust

notebooklm

Use this skill to query your Google NotebookLM notebooks directly from Claude Code for source-grounded, citation-backed answers from Gemini. Browser automation, library management, persistent auth. Drastically reduced hallucinations through document-only responses.

Skill与智能体开发
★ 2.0k