smart_prompt 0.4.4 → 0.5.0

data/examples/monitoring_example.rb

@@ -0,0 +1,121 @@
+#!/usr/bin/env ruby
+# Example demonstrating the monitoring and logging features of HistoryManager
+
+require_relative '../lib/smart_prompt'
+require 'logger'
+
+# Set up logger with INFO level
+SmartPrompt.logger = Logger.new($stdout)
+SmartPrompt.logger.level = Logger::INFO
+SmartPrompt.logger.formatter = proc do |severity, datetime, progname, msg|
+  "[#{datetime.strftime('%H:%M:%S')}] #{severity}: #{msg}\n"
+end
+
+puts "=" * 80
+puts "History Manager Monitoring Example"
+puts "=" * 80
+puts
+
+# Create a HistoryManager with monitoring enabled
+config = {
+  cache_size: 5,
+  session_defaults: {
+    max_messages: 10,
+    max_tokens: 1000
+  },
+  persistence: {
+    enabled: true,
+    storage_path: "./history_data_example",
+    async: false
+  },
+  monitoring: {
+    enabled: true,
+    log_level: :info
+  }
+}
+
+manager = SmartPrompt::HistoryManager.new(config)
+
+puts "\n--- Creating Sessions and Adding Messages ---\n"
+
+# Create first session
+manager.add_message("user_123", { role: "system", content: "You are a helpful assistant." })
+manager.add_message("user_123", { role: "user", content: "What is machine learning?" })
+manager.add_message("user_123", { role: "assistant", content: "Machine learning is a subset of AI..." })
+
+# Create second session
+manager.add_message("user_456", { role: "user", content: "Hello!" })
+manager.add_message("user_456", { role: "assistant", content: "Hi! How can I help you?" })
+
+puts "\n--- Getting Session Statistics ---\n"
+
+# Get statistics for a specific session
+session_stats = manager.get_stats("user_123")
+puts "Session user_123 statistics:"
+puts "  Messages: #{session_stats[:message_count]}"
+puts "  Tokens: #{session_stats[:total_tokens]}"
+puts "  Created: #{session_stats[:created_at]}"
+puts
+
+# Get system-wide statistics
+system_stats = manager.get_stats
+puts "System-wide statistics:"
+puts "  Active sessions: #{system_stats[:active_sessions]}"
+puts "  Total messages: #{system_stats[:total_messages]}"
+puts "  Total tokens: #{system_stats[:total_tokens]}"
+puts "  Messages per session (avg): #{system_stats[:messages_per_session_avg].round(2)}"
+puts "  Tokens per message (avg): #{system_stats[:tokens_per_message_avg].round(2)}"
+puts "  Cache hits: #{system_stats[:cache_hits]}"
+puts "  Cache misses: #{system_stats[:cache_misses]}"
+puts "  Cache hit rate: #{(system_stats[:cache_hit_rate] * 100).round(2)}%"
+puts
+
+puts "\n--- Exporting Metrics ---\n"
+
+# Export metrics in different formats
+puts "Prometheus format (first 10 lines):"
+prometheus_metrics = manager.export_metrics(format: :prometheus)
+puts prometheus_metrics.lines.first(10).join
+
+puts "\nJSON format:"
+json_metrics = manager.export_metrics(format: :json)
+require 'json'
+metrics_hash = JSON.parse(json_metrics)
+puts JSON.pretty_generate(metrics_hash.slice(
+  'active_sessions', 
+  'total_messages', 
+  'cache_hit_rate',
+  'messages_per_session_avg'
+))
+
+puts "\n--- Retrieving Context ---\n"
+
+# Retrieve context with token limit
+context = manager.get_context("user_123", 500)
+puts "Retrieved #{context.count} messages from user_123 (within 500 token limit)"
+
+puts "\n--- Searching Messages ---\n"
+
+# Search for messages
+results = manager.search_messages("user_123", "machine learning")
+puts "Found #{results.count} messages containing 'machine learning'"
+
+puts "\n--- Clearing Session ---\n"
+
+# Clear a session (keeping system messages)
+manager.clear_session("user_456", keep_system_messages: true)
+
+puts "\n--- Final Statistics ---\n"
+
+final_stats = manager.get_stats
+puts "Final system statistics:"
+puts "  Active sessions: #{final_stats[:active_sessions]}"
+puts "  Total messages: #{final_stats[:total_messages]}"
+puts "  Sessions created: #{final_stats[:sessions_created]}"
+puts "  Sessions deleted: #{final_stats[:sessions_deleted]}"
+puts "  Messages added: #{final_stats[:messages_added]}"
+puts "  Context retrievals: #{final_stats[:context_retrievals]}"
+
+# Cleanup
+manager.shutdown
+puts "\n--- Manager Shutdown Complete ---\n"

data/examples/multimodal_example.rb

@@ -0,0 +1,63 @@
+# Multimodal Example for SmartPrompt
+# This example demonstrates how to use the new MultimodalAdapter
+
+require_relative '../lib/smart_prompt'
+
+# Configuration for multimodal capabilities
+config = {
+  "adapters" => {
+    "multimodal" => "MultimodalAdapter"
+  },
+  "llms" => {
+    "qwen_vl" => {
+      "adapter" => "multimodal",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "default_model" => "Qwen/Qwen2.5-VL-7B-Instruct"
+    }
+  },
+  "default_llm" => "qwen_vl",
+  "template_path" => "./templates",
+  "worker_path" => "./workers",
+  "logger_file" => "./logs/smart_prompt.log"
+}
+
+# Write config to file
+File.write('multimodal_config.yml', config.to_yaml)
+
+# Initialize engine
+engine = SmartPrompt::Engine.new('multimodal_config.yml')
+
+# Example 1: Simple image analysis
+puts "=== Example 1: Image Analysis ==="
+result = engine.call_worker(:image_analyzer, {
+  image_url: "https://example.com/image.jpg",
+  question: "描述这张图片中的内容"
+})
+puts "Image Analysis Result: #{result}"
+
+# Example 2: Video analysis
+puts "\n=== Example 2: Video Analysis ==="
+result = engine.call_worker(:video_analyzer, {
+  video_url: "https://example.com/video.mp4",
+  question: "这个视频的主要内容是什么？",
+  max_frames: 15,
+  fps: 2
+})
+puts "Video Analysis Result: #{result}"
+
+# Example 3: Multiple images comparison
+puts "\n=== Example 3: Multiple Images Comparison ==="
+result = engine.call_worker(:multi_image_analyzer, {
+  image_urls: [
+    "https://example.com/image1.jpg",
+    "https://example.com/image2.jpg"
+  ],
+  question: "比较这两张图片的相似之处和不同之处"
+})
+puts "Multi-Image Analysis Result: #{result}"
+
+puts "\n=== All examples completed successfully ==="
+
+# Clean up
+File.delete('multimodal_config.yml') if File.exist?('multimodal_config.yml')

data/examples/relevance_based_strategy_example.rb

@@ -0,0 +1,87 @@
+#!/usr/bin/env ruby
+# Example demonstrating the RelevanceBasedStrategy
+
+require_relative '../lib/smart_prompt'
+
+# Create a session with some conversation history
+session = SmartPrompt::Session.new("demo_session", {})
+
+# Add a diverse set of messages
+session.add_message(role: "system", content: "You are a helpful AI assistant")
+session.add_message(role: "user", content: "Tell me about machine learning")
+session.add_message(role: "assistant", content: "Machine learning is a subset of artificial intelligence that enables systems to learn from data")
+session.add_message(role: "user", content: "What are your favorite animals?")
+session.add_message(role: "assistant", content: "I don't have personal preferences, but many people love cats and dogs")
+session.add_message(role: "user", content: "How does deep learning work?")
+session.add_message(role: "assistant", content: "Deep learning uses neural networks with multiple layers to learn hierarchical representations")
+session.add_message(role: "user", content: "Tell me a joke")
+session.add_message(role: "assistant", content: "Why did the programmer quit? Because they didn't get arrays!")
+session.add_message(role: "user", content: "What is supervised learning?")
+session.add_message(role: "assistant", content: "Supervised learning is when you train a model with labeled data")
+
+puts "=" * 80
+puts "RelevanceBasedStrategy Example"
+puts "=" * 80
+puts
+
+# Create the strategy
+strategy = SmartPrompt::RelevanceBasedStrategy.new(
+  top_k: 5,
+  recency_weight: 0.3,
+  relevance_weight: 0.7
+)
+
+# Current message is about neural networks
+current_message = SmartPrompt::Message.new(
+  role: "user",
+  content: "Can you explain more about neural networks and how they relate to machine learning?"
+)
+
+puts "Current message: #{current_message.content}"
+puts
+puts "Total messages in session: #{session.message_count}"
+puts
+
+# Select relevant messages
+messages = session.get_messages
+selected = strategy.select_messages(messages, nil, current_message)
+
+puts "Selected #{selected.length} most relevant messages:"
+puts "-" * 80
+selected.each_with_index do |msg, idx|
+  puts "#{idx + 1}. [#{msg.role}] #{msg.content}"
+end
+puts
+
+# Demonstrate with token limit
+puts "=" * 80
+puts "With Token Limit (100 tokens)"
+puts "=" * 80
+selected_limited = strategy.select_messages(messages, 100, current_message)
+total_tokens = selected_limited.sum { |m| m.token_count || 0 }
+
+puts "Selected #{selected_limited.length} messages (#{total_tokens} tokens):"
+puts "-" * 80
+selected_limited.each_with_index do |msg, idx|
+  tokens = msg.token_count || 0
+  puts "#{idx + 1}. [#{msg.role}] (#{tokens} tokens) #{msg.content[0..60]}..."
+end
+puts
+
+# Show compression recommendation
+puts "=" * 80
+puts "Compression Recommendation"
+puts "=" * 80
+should_compress = strategy.should_compress?(session)
+puts "Should compress? #{should_compress}"
+puts "Reason: Session has #{session.message_count} messages, threshold is #{5 * 3} messages"
+puts
+
+puts "=" * 80
+puts "Strategy Configuration"
+puts "=" * 80
+puts "Top-k: 5"
+puts "Recency weight: 0.3"
+puts "Relevance weight: 0.7"
+puts "Embedding service: Not configured (using keyword similarity)"
+puts "=" * 80

data/examples/sensenova_example.rb

@@ -0,0 +1,129 @@
+# SenseNova (商汤 日日新) Example for SmartPrompt
+#
+# Demonstrates all four SenseNova model categories through one SenseNovaAdapter:
+#   1. 商量 文本对话  (chat)        — sync + streaming
+#   2. 商量 图文多模态 (vision)
+#   3. Cupido 向量模型 (embeddings)
+#   4. 秒画 文生图    (text-to-image)
+#
+# Requires a valid SenseNova API key in the SENSENOVA_API_KEY environment variable
+# (get one at https://platform.sensenova.cn/console) and the relevant models enabled.
+
+require_relative "../lib/smart_prompt"
+
+api_key = ENV["SENSENOVA_API_KEY"]
+
+config = {
+  "adapters" => {
+    "sensenova" => "SenseNovaAdapter",
+  },
+  "llms" => {
+    "sensechat" => {
+      "adapter" => "sensenova",
+      "url" => "https://token.sensenova.cn/v1",
+      "api_key" => api_key,
+      "model" => "sensenova-6.7-flash-lite",
+      "temperature" => 0.7,
+    },
+    "sensevision" => {
+      "adapter" => "sensenova",
+      "url" => "https://token.sensenova.cn/v1",
+      "api_key" => api_key,
+      "model" => "sensenova-6.7-flash-lite",
+    },
+    "senseembedding" => {
+      "adapter" => "sensenova",
+      "url" => "https://api.sensenova.cn/compatible-mode/v2",
+      "embeddings_url" => "https://api.sensenova.cn/v1/llm/embeddings",
+      "api_key" => api_key,
+      "model" => "Cupido",
+    },
+    "senseimage" => {
+      "adapter" => "sensenova",
+      "url" => "https://token.sensenova.cn/v1",
+      "image_url" => "https://token.sensenova.cn/v1/images/generations",
+      "api_key" => api_key,
+      "model" => "sensenova-u1-fast",
+    },
+  },
+  "default_llm" => "sensechat",
+  "template_path" => "./templates",
+  "worker_path" => "./workers",
+  "logger_file" => "./logs/smart_prompt.log",
+}
+
+File.write("sensenova_config.yml", config.to_yaml)
+engine = SmartPrompt::Engine.new("sensenova_config.yml")
+
+puts "=== SmartPrompt SenseNova Demo ==="
+unless api_key
+  puts "Note: SENSENOVA_API_KEY is not set — the API calls below will fail at the network layer."
+end
+
+# 1. Chat (sync)
+puts "\n=== Example 1: 商量 文本对话 (sync) ==="
+begin
+  result = engine.call_worker(:sensenova_chat, { prompt: "用一句话介绍商汤日日新大模型。" })
+  puts "Reply: #{result}"
+rescue => e
+  puts "Error: #{e.message}"
+end
+
+# 2. Chat (streaming) — tokens are printed as they arrive.
+puts "\n=== Example 2: 商量 文本对话 (streaming) ==="
+begin
+  engine.call_worker_by_stream(:sensenova_chat, { prompt: "写两句关于春天的诗。" }) do |chunk, _|
+    if (delta = chunk.dig("choices", 0, "delta", "content"))
+      print delta
+    end
+  end
+  puts
+rescue => e
+  puts "Error: #{e.message}"
+end
+
+# 3. Multimodal vision
+puts "\n=== Example 3: 商量 图文多模态 ==="
+begin
+  result = engine.call_worker(:sensenova_vision, {
+    image_url: "https://img0.baidu.com/it/u=3775751201,1094020238&fm=253&fmt=auto&app=138&f=JPEG?w=500&h=615",
+    question: "图片里有什么？",
+  })
+  puts "Vision result: #{result}"
+rescue => e
+  puts "Error: #{e.message}"
+end
+
+# 4. Embeddings (Cupido)
+puts "\n=== Example 4: Cupido 向量模型 ==="
+begin
+  vector = engine.call_worker(:sensenova_embed, { text: "商汤日日新大模型", length: 1024 })
+  puts "Embedding dim: #{vector.is_a?(Array) ? vector.size : vector} (first 5: #{vector.first(5) rescue vector})"
+rescue => e
+  puts "Error: #{e.message}"
+end
+
+# 5. Text-to-image (秒画)
+puts "\n=== Example 5: 秒画 文生图 ==="
+begin
+  result = engine.call_worker(:sensenova_image, {
+    prompt: "一只在书房里读书的可爱机器人，温暖的光线，数字插画",
+    size: "2048x2048",
+    save_to_file: true,
+    output_dir: "./generated_images",
+    filename_prefix: "sensenova_robot",
+  })
+  if result.is_a?(Hash) && result[:images]
+    puts "Generated #{result[:images].size} image(s)"
+    puts "First image URL: #{result[:images].first[:url]}"
+    puts "Saved files: #{result[:saved_files]}"
+  else
+    puts "Result: #{result}"
+  end
+rescue => e
+  puts "Error (image endpoint may need a live key to confirm): #{e.message}"
+end
+
+puts "\n=== All examples completed ==="
+
+File.delete("sensenova_config.yml") if File.exist?("sensenova_config.yml")

data/examples/stt_example.rb

@@ -0,0 +1,287 @@
+# STT Example for SmartPrompt
+# This example demonstrates how to use the new STTAdapter
+
+require_relative '../lib/smart_prompt'
+
+# Configuration for STT capabilities
+config = {
+  "adapters" => {
+    "multimodal" => "MultimodalAdapter",
+    "image_generation" => "ImageGenerationAdapter",
+    "video_generation" => "VideoGenerationAdapter",
+    "tts" => "TTSAdapter",
+    "stt" => "STTAdapter"
+  },
+  "llms" => {
+    "qwen_vl" => {
+      "adapter" => "multimodal",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "model" => "Qwen/Qwen2.5-VL-7B-Instruct"
+    },
+    "image_gen" => {
+      "adapter" => "image_generation",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "model" => "stabilityai/stable-diffusion-xl-base-1.0"
+    },
+    "video_gen" => {
+      "adapter" => "video_generation",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "model" => "Wan-AI/Wan2.2-T2V-A14B"
+    },
+    "tts_service" => {
+      "adapter" => "tts",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "model" => "FunAudioLLM/CosyVoice2-0.5B"
+    },
+    "stt_service" => {
+      "adapter" => "stt",
+      "url" => "https://api.siliconflow.cn/v1/",
+      "api_key" => ENV["SILICONFLOW_API_KEY"],
+      "model" => "FunAudioLLM/CosyVoice2-0.5B"
+    }
+  },
+  "default_llm" => "qwen_vl",
+  "template_path" => "./templates",
+  "worker_path" => "./workers",
+  "logger_file" => "./logs/smart_prompt.log"
+}
+
+# Write config to file
+File.write('stt_config.yml', config.to_yaml)
+
+# Initialize engine
+engine = SmartPrompt::Engine.new('stt_config.yml')
+
+puts "=== SmartPrompt STT Demo ==="
+
+# Example 1: Basic speech-to-text transcription
+puts "\n=== Example 1: Basic STT Transcription ==="
+begin
+  # Note: This example requires an actual audio file
+  # Replace with a real audio file path for testing
+  audio_file_path = "./test_audio.wav"
+
+  if File.exist?(audio_file_path)
+    result = engine.call_worker(:stt_transcriber, {
+      audio_file: audio_file_path,
+      language: "zh",
+      response_format: "json"
+    })
+
+    puts "STT transcription successful!"
+    puts "Transcribed text: #{result[:transcription][:text]}"
+    puts "Language: #{result[:transcription][:language]}"
+    puts "Duration: #{result[:transcription][:duration]} seconds"
+    puts "File size: #{result[:transcription][:file_size]} bytes"
+  else
+    puts "Audio file not found: #{audio_file_path}"
+    puts "Please create a test audio file to run this example"
+  end
+
+rescue => e
+  puts "Error in STT transcription: #{e.message}"
+  puts "Note: This example requires a valid SILICONFLOW_API_KEY environment variable"
+end
+
+# Example 2: URL-based transcription
+puts "\n=== Example 2: URL-based STT Transcription ==="
+begin
+  # Note: Replace with a real audio URL for testing
+  audio_url = "https://example.com/audio.wav"
+
+  result = engine.call_worker(:stt_url_transcriber, {
+    audio_url: audio_url,
+    language: "en",
+    response_format: "text"
+  })
+
+  puts "URL-based STT transcription successful!"
+  puts "Transcribed text: #{result[:transcription][:text]}"
+  puts "Audio URL: #{result[:transcription][:audio_url]}"
+
+rescue => e
+  puts "Error in URL-based STT: #{e.message}"
+  puts "Note: This requires a valid audio URL"
+end
+
+# Example 3: Batch transcription
+puts "\n=== Example 3: Batch STT Processing ==="
+begin
+  # Note: Replace with real audio files for testing
+  audio_files = ["./audio1.wav", "./audio2.wav", "./audio3.wav"]
+  existing_files = audio_files.select { |f| File.exist?(f) }
+
+  if existing_files.any?
+    result = engine.call_worker(:batch_stt, {
+      audio_files: existing_files,
+      language: "zh"
+    })
+
+    puts "Batch STT processing successful!"
+    puts "Total files: #{result[:batch_result][:total_files]}"
+    puts "Successful: #{result[:batch_result][:successful]}"
+    puts "Failed: #{result[:batch_result][:failed]}"
+
+    result[:batch_result][:results].each do |file_result|
+      if file_result[:success]
+        puts "  - #{File.basename(file_result[:file])}: #{file_result[:transcription][:text].length} characters"
+      else
+        puts "  - #{File.basename(file_result[:file])}: ERROR - #{file_result[:error]}"
+      end
+    end
+  else
+    puts "No audio files found for batch processing"
+    puts "Please create test audio files to run this example"
+  end
+
+rescue => e
+  puts "Error in batch STT: #{e.message}"
+end
+
+# Example 4: Audio file information
+puts "\n=== Example 4: Audio File Information ==="
+begin
+  audio_file_path = "./test_audio.wav"
+
+  if File.exist?(audio_file_path)
+    result = engine.call_worker(:audio_info, {
+      audio_file: audio_file_path
+    })
+
+    puts "Audio file information retrieved!"
+    puts "File name: #{result[:audio_info][:file_name]}"
+    puts "File size: #{result[:audio_info][:file_size]} bytes"
+    puts "Format: #{result[:audio_info][:format]}"
+    puts "Estimated duration: #{result[:audio_info][:estimated_duration]} seconds"
+    puts "Supported: #{result[:audio_info][:supported]}"
+  else
+    puts "Audio file not found: #{audio_file_path}"
+  end
+
+rescue => e
+  puts "Error getting audio info: #{e.message}"
+end
+
+# Example 5: Language detection
+puts "\n=== Example 5: Language Detection ==="
+begin
+  # Test with Chinese text
+  result = engine.call_worker(:language_detector, {
+    text: "这是一个中文文本，用于语言检测演示。"
+  })
+
+  puts "Language detection successful!"
+  puts "Text: #{result[:text]}"
+  puts "Detected language: #{result[:detected_language]}"
+
+  # Test with English text
+  result_en = engine.call_worker(:language_detector, {
+    text: "This is an English text for language detection demonstration."
+  })
+
+  puts "English text detected as: #{result_en[:detected_language]}"
+
+rescue => e
+  puts "Error in language detection: #{e.message}"
+end
+
+# Example 6: Multi-language STT
+puts "\n=== Example 6: Multi-language STT ==="
+begin
+  audio_file_path = "./test_audio.wav"
+
+  if File.exist?(audio_file_path)
+    result = engine.call_worker(:multilingual_stt, {
+      audio_file: audio_file_path
+    })
+
+    puts "Multi-language STT successful!"
+    puts "Detected language: #{result[:detected_language]}"
+    puts "Initial transcription: #{result[:initial_transcription][:text]}"
+
+    if result[:improved_transcription]
+      puts "Improved transcription: #{result[:improved_transcription][:text]}"
+    end
+  else
+    puts "Audio file not found: #{audio_file_path}"
+  end
+
+rescue => e
+  puts "Error in multi-language STT: #{e.message}"
+end
+
+# Example 7: Format conversion
+puts "\n=== Example 7: STT Format Conversion ==="
+begin
+  audio_file_path = "./test_audio.wav"
+
+  if File.exist?(audio_file_path)
+    result = engine.call_worker(:stt_format_converter, {
+      audio_file: audio_file_path,
+      formats: ["json", "text", "srt", "vtt"]
+    })
+
+    puts "Format conversion successful!"
+    result[:format_results].each do |format, transcription|
+      puts "  - #{format.upcase}: #{transcription[:text].length} characters"
+    end
+  else
+    puts "Audio file not found: #{audio_file_path}"
+  end
+
+rescue => e
+  puts "Error in format conversion: #{e.message}"
+end
+
+# Example 8: Direct adapter usage
+puts "\n=== Example 8: Direct Adapter Usage ==="
+begin
+  # Get the adapter directly
+  adapter = engine.llms["stt_service"]
+
+  audio_file_path = "./test_audio.wav"
+
+  if File.exist?(audio_file_path)
+    # Transcribe audio directly
+    transcription_data = adapter.transcribe_audio(
+      audio_file_path,
+      language: "zh",
+      temperature: 0.0,
+      response_format: "json"
+    )
+
+    puts "Direct adapter usage successful!"
+    puts "Transcribed text: #{transcription_data[:text]}"
+    puts "Language: #{transcription_data[:language]}"
+    puts "Duration: #{transcription_data[:duration]} seconds"
+
+    # Get audio information
+    audio_info = adapter.get_audio_info(audio_file_path)
+    puts "Audio info - Format: #{audio_info[:format]}, Size: #{audio_info[:file_size]} bytes"
+
+    # Detect language
+    detected_language = adapter.detect_language(transcription_data[:text])
+    puts "Detected language: #{detected_language}"
+
+  else
+    puts "Audio file not found: #{audio_file_path}"
+  end
+
+rescue => e
+  puts "Error in direct adapter usage: #{e.message}"
+end
+
+puts "\n=== All examples completed ==="
+puts "\nImportant Notes:"
+puts "1. STT requires valid SILICONFLOW_API_KEY environment variable"
+puts "2. Audio files must be in supported formats (mp3, wav, webm, etc.)"
+puts "3. Maximum file size: 25MB"
+puts "4. Supported languages: Chinese, English, Japanese, Korean"
+puts "5. Response formats: json, text, srt, vtt"
+
+# Clean up
+File.delete('stt_config.yml') if File.exist?('stt_config.yml')