🎤 Channel TTS

Production-Ready Korean Text-to-Speech API

Channel TTS는 채널코퍼레이션에서 개발한 엔터프라이즈급 한국어 음성 합성 시스템입니다.
실시간 대화형 서비스에 최적화된 낮은 레이턴시와 자연스러운 음성 품질을 제공합니다.

⚡ Key Features

⚡

Real-time Streaming

TTFA ~140-240ms
RTF ~0.2-0.3x

🎚️

Adaptive Latency

5-level optimization
(0-4 configurable)

🎵

Multi-format Support

PCM 16/24/44.1/48kHz
WAV output

🇰🇷

Korean-optimized

Advanced normalization
Natural prosody

🚀 Quick Start

Streaming (Real-time)

# 기본 사용 (최고 품질, optimize_streaming_latency=0)
curl -X POST "https://ch-tts-streaming-demo.channel.io/v1/external/text-to-speech/hana/stream?optimize_streaming_latency=0" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $API_TOKEN" \
  -d '{"text":"안녕하세요. 채널톡입니다.","output_format":"pcm_24000"}' \
  | ffplay -nodisp -autoexit -f s16le -ar 24000 -i -

Batch (File generation)

curl -X POST "https://ch-tts-streaming-demo.channel.io/v1/external/text-to-speech/hana" \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $API_TOKEN" \
  -d '{"text":"안녕하세요","output_format":"wav_24000"}' \
  -o output.wav

📊 Performance Metrics

Metric	Value	Description
TTFB	~140-240ms	Time to First Byte (첫 바이트까지 걸리는 시간)
TTFA	~140-240ms	Time to First Audio (첫 오디오 청크까지 걸리는 시간)
RTF	~0.2-0.3x	Real-Time Factor (실시간 대비 3-5배 빠른 생성 속도)

🎯 Optimization Levels

Configure latency vs quality trade-off with optimize_streaming_latency parameter (0-4):

Level	Quality	Speed	Use Case
0	⭐⭐⭐⭐⭐	Slowest	High-quality voice generation
1	⭐⭐⭐⭐	Slow	Balanced quality and speed
2	⭐⭐⭐	Medium	General conversation
3	⭐⭐	Fast	Real-time interaction
4	⭐	Fastest	Ultra-low latency required