@hienlh/ppm 0.9.81 → 0.9.83

package/.opencode/skills/elevenlabs/references/conversational-ai-agents-guide.md

@@ -1,168 +0,0 @@
-# Conversational AI Agents Guide
-
-Build real-time voice agents with ElevenLabs Agents Platform.
-
-## SDK Installation
-
-```bash
-# Python (with audio support)
-pip install "elevenlabs[pyaudio]"
-
-# JavaScript/Node.js
-npm install @11labs/client
-```
-
-## Python Quick Start
-
-```python
-from elevenlabs import ElevenLabs
-from elevenlabs.conversational_ai.conversation import Conversation
-
-# Initialize client
-client = ElevenLabs(api_key="YOUR_API_KEY")
-
-# Start conversation (uses default system audio)
-conversation = Conversation.start_session(
-    agent_id="YOUR_AGENT_ID",
-    on_message=lambda msg: print(f"Agent: {msg}"),
-    on_user_message=lambda msg: print(f"User: {msg}"),
-    on_error=lambda e: print(f"Error: {e}")
-)
-
-# The conversation runs until stopped
-conversation.wait_for_session_end()
-```
-
-## JavaScript Quick Start
-
-```javascript
-import { Conversation } from "@11labs/client";
-
-const conversation = await Conversation.startSession({
-  agentId: "YOUR_AGENT_ID",
-  onMessage: (message) => console.log("Agent:", message),
-  onUserMessage: (message) => console.log("User:", message),
-  onConnect: () => console.log("Connected"),
-  onDisconnect: () => console.log("Disconnected"),
-});
-
-// End session
-conversation.endSession();
-```
-
-## WebSocket Streaming
-
-For custom audio handling:
-
-```python
-import websocket
-import json
-
-def on_message(ws, message):
-    data = json.loads(message)
-    if data["type"] == "audio":
-        # Handle audio chunk
-        audio_bytes = base64.b64decode(data["audio"])
-
-ws = websocket.WebSocketApp(
-    "wss://api.elevenlabs.io/v1/convai/conversation",
-    header={"xi-api-key": "YOUR_API_KEY"},
-    on_message=on_message
-)
-
-# Send user audio
-ws.send(json.dumps({
-    "type": "audio",
-    "audio": base64.b64encode(audio_chunk).decode()
-}))
-```
-
-## Phone Integration
-
-### Twilio Setup
-
-1. Create agent in ElevenLabs dashboard
-2. Get webhook URL from agent settings
-3. Configure Twilio number to forward to webhook
-
-```python
-# Twilio webhook handler
-from flask import Flask, request
-app = Flask(__name__)
-
-@app.route("/voice", methods=["POST"])
-def voice():
-    return f"""
-    <Response>
-        <Connect>
-            <Stream url="wss://api.elevenlabs.io/v1/convai/twilio/{AGENT_ID}" />
-        </Connect>
-    </Response>
-    """
-```
-
-### Supported Providers
-- Twilio
-- Vonage
-- Telnyx
-- Plivo
-- Genesys
-- Generic SIP
-
-## Agent Configuration
-
-### Knowledge Base (RAG)
-Upload documents for agent to reference:
-
-```python
-client.agents.knowledge_base.add(
-    agent_id="YOUR_AGENT_ID",
-    files=["product_docs.pdf", "faq.txt"]
-)
-```
-
-### Custom Tools
-
-Define tools agent can invoke:
-
-```python
-tools = [
-    {
-        "name": "get_weather",
-        "description": "Get current weather for a city",
-        "parameters": {
-            "type": "object",
-            "properties": {
-                "city": {"type": "string"}
-            }
-        }
-    }
-]
-```
-
-### Voice Selection
-
-Use low-latency voices for agents:
-- Flash v2.5 recommended (~75ms)
-- Turbo v2.5 for higher quality (~250ms)
-
-## Best Practices
-
-1. **Latency**: Use Flash models + streaming
-2. **Interruption Handling**: Configure turn detection sensitivity
-3. **Error Recovery**: Implement reconnection logic
-4. **Testing**: Use agent test mode before production
-5. **Monitoring**: Track conversation metrics in dashboard
-
-## Concurrency Limits
-
-| Plan | Agent Connections |
-|------|------------------|
-| Free | 2 |
-| Starter | 3 |
-| Creator | 5 |
-| Pro | 10 |
-| Scale | 15 |
-| Enterprise | Custom |
-
-~5 concurrent connections supports ~100 simultaneous audio broadcasts.

package/.opencode/skills/elevenlabs/references/sound-effects-and-music-generation-guide.md

@@ -1,131 +0,0 @@
-# Sound Effects & Music Generation Guide
-
-## Sound Effects API
-
-### Endpoint
-```
-POST https://api.elevenlabs.io/v1/sound-generation
-```
-
-### Limits
-- Max duration: 30 seconds
-- Output format: MP3 (48kHz WAV available for non-looping)
-- Looping: Supported for seamless repeating audio
-
-### Prompt Tips
-
-**Be Descriptive:**
-```
-A heavy wooden door creaking open slowly in an old mansion,
-with echoing hinges and dust settling
-```
-
-**Specify Duration:**
-```
-5 second thunderclap with rolling rumble
-```
-
-**Request Loops:**
-```
-Looping ambient forest sounds with birds chirping and wind through leaves
-```
-
-### Example Prompts
-
-| Category | Prompt |
-|----------|--------|
-| **Cinematic** | "Deep, ominous drone building tension, 10 seconds" |
-| **UI** | "Short, crisp notification ping, friendly tone" |
-| **Game** | "8-bit coin collection sound, retro arcade style" |
-| **Foley** | "Footsteps on gravel, slow walking pace, outdoor" |
-| **Ambient** | "Looping coffee shop ambience with quiet chatter and espresso machine" |
-
-### Python Example
-
-```python
-from elevenlabs import ElevenLabs
-
-client = ElevenLabs(api_key="YOUR_API_KEY")
-
-audio = client.sound_effects.generate(
-    text="Thunder rumbling in the distance with light rain",
-    duration_seconds=10.0
-)
-
-with open("thunder.mp3", "wb") as f:
-    for chunk in audio:
-        f.write(chunk)
-```
-
-## Music Generation (Eleven Music)
-
-### Requirements
-- Paid plan required
-- Duration: 10 seconds to 5 minutes
-- Commercially licensed (additional license for ads/film/TV/games)
-
-### Endpoint
-```
-POST https://api.elevenlabs.io/v1/music/compose
-```
-
-### Basic Prompt
-
-```python
-music = client.music.compose(
-    prompt="Upbeat electronic track with driving beat,
-            perfect for tech product video",
-    duration_seconds=60
-)
-```
-
-### Composition Plans (Advanced)
-
-For granular control, use a composition plan:
-
-```python
-plan = {
-    "sections": [
-        {
-            "name": "intro",
-            "duration": 15,
-            "style": "ambient, building"
-        },
-        {
-            "name": "main",
-            "duration": 30,
-            "style": "energetic, full instrumentation"
-        },
-        {
-            "name": "outro",
-            "duration": 15,
-            "style": "fadeout, ambient"
-        }
-    ]
-}
-
-music = client.music.compose_detailed(plan=plan)
-```
-
-### Prompt Examples
-
-| Use Case | Prompt |
-|----------|--------|
-| **Podcast Intro** | "30 second upbeat acoustic guitar melody, warm and welcoming" |
-| **Game Background** | "Looping orchestral adventure theme, epic fantasy style" |
-| **Corporate Video** | "Professional, inspiring corporate track with piano and strings" |
-| **Social Media** | "15 second trendy lo-fi hip hop beat, chill vibes" |
-
-### Tips
-
-1. **Specify genre clearly**: "jazz", "electronic", "orchestral"
-2. **Include mood**: "energetic", "melancholic", "triumphant"
-3. **Mention instruments** if specific ones needed
-4. **Request vocals or instrumental**: Default is usually instrumental
-5. **Duration affects structure**: Longer tracks need more detailed prompts
-
-## Billing
-
-- Sound effects: Per generation
-- Music: Per generation
-- Both count against monthly quota on paid plans

package/.opencode/skills/elevenlabs/references/text-to-speech-guide.md

@@ -1,132 +0,0 @@
-# Text-to-Speech Guide
-
-## API Endpoint
-
-```
-POST https://api.elevenlabs.io/v1/text-to-speech/{voice_id}
-```
-
-## Voice Settings
-
-| Setting | Range | Default | Purpose |
-|---------|-------|---------|---------|
-| `stability` | 0-1 | 0.5 | Higher = more consistent, lower = more expressive |
-| `similarity_boost` | 0-1 | 0.75 | How closely to match original voice |
-| `style` | 0-1 | 0 | Expressiveness (Multilingual v2 only) |
-| `speed` | 0.7-1.2 | 1.0 | Playback speed multiplier |
-
-## SSML Tags (English models only)
-
-### Pauses
-```xml
-<break time="1.5s" />
-```
-- Max 3 seconds per break
-- Too many breaks cause instability
-- Eleven v3 doesn't support SSML; use punctuation instead
-
-### Pronunciation (CMU Arpabet - recommended)
-```xml
-<phoneme alphabet="cmu-arpabet" ph="T OW M EY T OW">tomato</phoneme>
-```
-
-### Pronunciation (IPA)
-```xml
-<phoneme alphabet="ipa" ph="təˈmeɪtoʊ">tomato</phoneme>
-```
-
-**Stress markers are critical** - include them for accurate pronunciation.
-
-### Alternative Pause Techniques
-- Dash `-` or em-dash `—` for short pauses
-- Ellipsis `...` for hesitant tone
-- Comma for natural breath pauses
-
-## Emotional Control
-
-Models interpret emotion from context:
-- Add descriptive phrases: "she said excitedly"
-- Use punctuation: `!` for excitement, `?` for questioning
-- Wrap in context: `[whispering]` or `[angrily]`
-
-Match emotion tags to voice training data - professional voices may not respond to playful tags.
-
-## Text Normalization
-
-Expand for best results:
-- `01/02/2023` → `January second, twenty twenty-three`
-- `$50` → `fifty dollars`
-- `Dr.` → `Doctor`
-
-Set `apply_text_normalization: "on"` in API for auto-normalization.
-
-## Streaming
-
-Use streaming endpoint for reduced time-to-first-byte:
-```
-POST https://api.elevenlabs.io/v1/text-to-speech/{voice_id}/stream
-```
-
-**Chunk sizes:**
-- Minimum: 50 characters
-- Recommended: 200-400 characters
-- Maximum: 1,000 characters
-
-## Output Formats
-
-| Format | Quality | Size |
-|--------|---------|------|
-| `mp3_44100_128` | High | Medium |
-| `mp3_22050_32` | Low | Small |
-| `pcm_16000` | Raw | Large |
-| `ulaw_8000` | Telephony | Small |
-
-## Global API for Lower Latency
-
-Replace `api.elevenlabs.io` with `api-global-preview.elevenlabs.io` for:
-- Flash v2.5
-- Turbo v2.5
-
-Not recommended for Multilingual v2 (may increase latency).
-
-## Python Example
-
-```python
-from elevenlabs import ElevenLabs
-
-client = ElevenLabs(api_key="YOUR_API_KEY")
-
-audio = client.text_to_speech.convert(
-    voice_id="JBFqnCBsd6RMkjVDRZzb",
-    text="Hello! <break time='0.5s'/> How are you?",
-    model_id="eleven_multilingual_v2",
-    voice_settings={
-        "stability": 0.5,
-        "similarity_boost": 0.75,
-        "style": 0.3,
-        "speed": 1.0
-    }
-)
-
-# Save to file
-with open("output.mp3", "wb") as f:
-    for chunk in audio:
-        f.write(chunk)
-```
-
-## JavaScript Example
-
-```javascript
-import { ElevenLabsClient } from "@elevenlabs/elevenlabs-js";
-
-const client = new ElevenLabsClient({ apiKey: "YOUR_API_KEY" });
-
-const audio = await client.textToSpeech.convert("JBFqnCBsd6RMkjVDRZzb", {
-  text: "Hello world!",
-  model_id: "eleven_multilingual_v2",
-  voice_settings: {
-    stability: 0.5,
-    similarity_boost: 0.75,
-  },
-});
-```

package/.opencode/skills/elevenlabs/references/voice-cloning-guide.md

@@ -1,94 +0,0 @@
-# Voice Cloning Guide
-
-## Two Cloning Methods
-
-| Method | Audio Required | Training Time | Quality |
-|--------|---------------|---------------|---------|
-| **Instant (IVC)** | 1-2 minutes | Instant | Good |
-| **Professional (PVC)** | 30+ minutes | Hours | Excellent |
-
-## Instant Voice Cloning
-
-Fast, easy cloning for quick projects.
-
-### Requirements
-- 1-2 minutes of clear audio
-- No background noise, reverb, or artifacts
-- Consistent speaking style throughout
-- **Don't exceed 3 minutes** - can reduce quality
-
-### API Usage
-
-```python
-from elevenlabs import ElevenLabs
-
-client = ElevenLabs(api_key="YOUR_API_KEY")
-
-voice = client.voices.add(
-    name="My Voice Clone",
-    files=["sample1.mp3", "sample2.mp3"],
-    description="Voice clone for narration"
-)
-
-print(f"Voice ID: {voice.voice_id}")
-```
-
-## Professional Voice Cloning
-
-High-fidelity cloning for production use.
-
-### Requirements
-- Minimum 30 minutes, ideally 2+ hours
-- Recording style should match intended output
-- Clean, studio-quality audio
-- Consistent tone and pacing throughout
-
-### Recording Guidelines
-
-**Microphone Setup:**
-- Distance: ~20cm (7-8 inches) from mic
-- Use pop filter between mouth and mic
-- Quiet room with minimal reverb
-
-**Speaking Style:**
-- Stay consistent - don't mix animated and subdued
-- Keep same accent throughout
-- Insert 1-1.5s silences between paragraphs
-- Shorter pauses between sentences
-- Avoid throat clearing, vocal fry (unless intentional)
-
-### What NOT to Do
-- Mix different energy levels
-- Switch accents mid-recording
-- Include long silences or dead air
-- Record in noisy environments
-
-## Voice Quality Tips
-
-1. **Sample Diversity**: Include various emotions and pacing in samples
-2. **Audio Quality**: 44.1kHz, 16-bit minimum
-3. **Format**: WAV or high-bitrate MP3
-4. **Content**: Natural speech, not reading robotically
-
-## Managing Cloned Voices
-
-```python
-# List all voices
-voices = client.voices.get_all()
-for voice in voices.voices:
-    print(f"{voice.name}: {voice.voice_id}")
-
-# Delete a voice
-client.voices.delete(voice_id="voice_id_here")
-
-# Get voice details
-voice = client.voices.get(voice_id="voice_id_here")
-print(voice.settings)
-```
-
-## Sharing & Permissions
-
-- Cloned voices are private by default
-- Can be shared via Voice Library
-- Enterprise: Cross-workspace sharing available
-- Always verify you have rights to clone a voice

package/.opencode/skills/elevenlabs/references/voice-design-prompting-guide.md

@@ -1,96 +0,0 @@
-# Voice Design Prompting Guide
-
-Create custom voices from text descriptions when Voice Library doesn't have what you need.
-
-## Prompt Structure
-
-### Five Core Attributes
-
-1. **Age**: adolescent, young adult, middle-aged, elderly
-2. **Gender**: male, female, neutral, or pitch descriptors
-3. **Tone/Timbre**: deep, smooth, gravelly, warm, breathy, tinny
-4. **Accent**: use "thick" or "slight" + specific accent
-5. **Pacing**: fast, slow, deliberate, conversational, rhythmic
-
-## Prompt Quality Spectrum
-
-### Detailed Prompts (Most Accurate)
-```
-A high-energy female sports commentator with a thick British accent,
-passionately delivering play-by-play coverage of a football match
-at a very quick pace.
-```
-
-### Simple Prompts (Good for Neutral)
-```
-A calm male narrator
-```
-
-## Example Prompts by Use Case
-
-### Audiobook Narrator
-```
-A warm, middle-aged male voice with a slight British accent.
-Calm, measured pacing perfect for long-form narration.
-Studio-quality recording with rich, deep tones.
-```
-
-### Energetic Podcast Host
-```
-A young, enthusiastic female voice with an American accent.
-Quick, dynamic delivery with natural energy and warmth.
-Conversational tone, perfect audio quality.
-```
-
-### Documentary Narrator
-```
-An authoritative, deep male voice. Middle-aged with gravelly undertones.
-Slow, deliberate pacing. Serious and informative tone.
-```
-
-### Character Voice (Pirate)
-```
-A gruff, weathered old man with a thick West Country English accent.
-Raspy voice, speaking slowly with dramatic pauses.
-Sounds like years at sea.
-```
-
-### Tech Product Demo
-```
-A friendly, professional young female voice. Slight uptalk pattern.
-Clear enunciation, moderate pace. Approachable and modern.
-Perfect audio quality.
-```
-
-## Quality Modifiers
-
-### For High Quality
-Add: "perfect audio quality" or "studio-quality recording"
-
-### For Creative Lo-Fi
-- "low-fidelity audio"
-- "sounds like a voicemail"
-- "muffled and distant"
-- "over a phone line"
-
-## Parameters
-
-| Parameter | Effect |
-|-----------|--------|
-| **Loudness** | Volume of generated voice |
-| **Guidance Scale** | 20-40% for most cases. Higher = stricter prompt adherence but may reduce quality |
-
-## Tips
-
-1. **Be specific**: "thick French accent" > "foreign accent"
-2. **Combine traits**: Layer multiple descriptors for unique voices
-3. **Match preview text**: Use text that complements your voice description
-4. **Longer previews**: Produce more stable, expressive results
-5. **Iterate**: Generate multiple versions and pick the best
-
-## What to Avoid
-
-- Vague terms: "exotic", "foreign", "interesting"
-- Contradictory descriptors in same prompt
-- Extremely specific celebrity impersonations
-- Overly long prompts (diminishing returns)

package/.opencode/skills/elevenlabs/requirements.txt

@@ -1 +0,0 @@
-elevenlabs>=1.0.0