docs: TTS_BATCH_SIZE
This commit is contained in:
parent
fa27fe561a
commit
124d7d110c
2 changed files with 39 additions and 0 deletions
|
|
@ -42,6 +42,12 @@
|
||||||
# Used only by the podcast feature
|
# Used only by the podcast feature
|
||||||
# ELEVENLABS_API_KEY=
|
# ELEVENLABS_API_KEY=
|
||||||
|
|
||||||
|
# TTS BATCH SIZE
|
||||||
|
# Controls concurrent TTS requests for podcast generation (default: 5)
|
||||||
|
# Lower values reduce provider load but increase generation time
|
||||||
|
# Recommended: OpenAI=5, ElevenLabs=2, Google=4, Custom=1
|
||||||
|
# TTS_BATCH_SIZE=2
|
||||||
|
|
||||||
# VOYAGE AI
|
# VOYAGE AI
|
||||||
# VOYAGE_API_KEY=
|
# VOYAGE_API_KEY=
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -197,11 +197,32 @@ Each speaker profile includes:
|
||||||
- **Batch Processing**: Automated generation workflows
|
- **Batch Processing**: Automated generation workflows
|
||||||
|
|
||||||
### Advanced Configurations
|
### Advanced Configurations
|
||||||
|
|
||||||
|
#### **Performance Tuning**
|
||||||
- **Segment Structure**: Custom conversation organization
|
- **Segment Structure**: Custom conversation organization
|
||||||
- **Timing Control**: Precise episode length management
|
- **Timing Control**: Precise episode length management
|
||||||
- **Topic Weighting**: Emphasis on specific content areas
|
- **Topic Weighting**: Emphasis on specific content areas
|
||||||
- **Personality Mixing**: Complex speaker interaction patterns
|
- **Personality Mixing**: Complex speaker interaction patterns
|
||||||
|
|
||||||
|
#### **TTS Concurrency Control**
|
||||||
|
Configure parallel audio generation to optimize performance and avoid provider rate limits:
|
||||||
|
|
||||||
|
```bash
|
||||||
|
# Environment variable configuration
|
||||||
|
export TTS_BATCH_SIZE=3 # Number of concurrent TTS requests (default: 5)
|
||||||
|
```
|
||||||
|
|
||||||
|
**Recommended Settings by Provider:**
|
||||||
|
- **OpenAI TTS**: `TTS_BATCH_SIZE=5` (default, handles high concurrency well)
|
||||||
|
- **ElevenLabs**: `TTS_BATCH_SIZE=2` (strict rate limits, reduce for stability)
|
||||||
|
- **Google TTS**: `TTS_BATCH_SIZE=4` (moderate concurrency tolerance)
|
||||||
|
- **Custom/Local TTS**: `TTS_BATCH_SIZE=1` (depends on hardware/setup)
|
||||||
|
|
||||||
|
**Performance Trade-offs:**
|
||||||
|
- **Higher values (4-5)**: Faster podcast generation, higher provider load
|
||||||
|
- **Lower values (1-2)**: Slower generation, more reliable for rate-limited providers
|
||||||
|
- **Optimal setting**: Balance between speed and provider stability
|
||||||
|
|
||||||
## 🛠️ Troubleshooting Common Issues
|
## 🛠️ Troubleshooting Common Issues
|
||||||
|
|
||||||
### Generation Failures
|
### Generation Failures
|
||||||
|
|
@ -215,6 +236,18 @@ Each speaker profile includes:
|
||||||
- **Solution**: Check API quotas and upgrade plans if needed
|
- **Solution**: Check API quotas and upgrade plans if needed
|
||||||
- **Prevention**: Monitor usage and set up billing alerts
|
- **Prevention**: Monitor usage and set up billing alerts
|
||||||
|
|
||||||
|
#### **TTS Concurrency Issues**
|
||||||
|
- **Problem**: TTS provider rate limiting or concurrent request failures
|
||||||
|
- **Solution**: Configure TTS batch size to reduce parallel audio generation
|
||||||
|
- **Environment Variable**: `TTS_BATCH_SIZE=2` (default: 5)
|
||||||
|
- **Usage**: Lower values reduce provider load but increase generation time
|
||||||
|
```bash
|
||||||
|
# Reduce concurrent TTS requests for providers with strict limits
|
||||||
|
export TTS_BATCH_SIZE=2
|
||||||
|
# or
|
||||||
|
export TTS_BATCH_SIZE=1 # Most conservative, slowest
|
||||||
|
```
|
||||||
|
|
||||||
#### **Voice Configuration Errors**
|
#### **Voice Configuration Errors**
|
||||||
- **Problem**: Specific voice not available or misconfigured
|
- **Problem**: Specific voice not available or misconfigured
|
||||||
- **Solution**: Verify TTS provider settings and voice availability
|
- **Solution**: Verify TTS provider settings and voice availability
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue