diff --git a/public/text-to-speech.html b/public/text-to-speech.html
index e0339d7..d247bea 100644
--- a/public/text-to-speech.html
+++ b/public/text-to-speech.html
@@ -108,7 +108,7 @@
- Zero API Keys: No registration, no tracking, no rate limits on your own infrastructure
- OpenAI-Compatible: Drop-in replacement - change one URL and you're running
- - Five TTS Engines: Qwen3-TTS (default, voice cloning), Piper (fast), XTTS (quality), Silero (CPU-friendly), Kokoro (lightweight)
+ - Five TTS Engines: Voice cloning (default), fast synthesis, HD cloning, multilingual CPU, and lightweight edge
- 42+ Built-In Cloned Voices: With 10-language support including Chinese, Japanese, Korean, and European languages
- Self-Hostable: Docker compose, Makefile-driven, runs on your hardware
- AGPL v3 Licensed: Keeps TTS libre forever - even network service users get source code
@@ -169,13 +169,17 @@ client.audio.speech.create(
Discover Voices
- Browse all available voices with engine metadata: speech.ai.unturf.com/v1/voices
+ Every voice has a name, a gender, and a personality. Browse the full catalog live â the API tells you exactly what's available and which engine powers each one.
+
+ â See all voices
curl https://speech.ai.unturf.com/v1/voices | jq .
Discover Models
- See all available TTS engines: speech.ai.unturf.com/v1/models
+ Five engines, each with different strengths. Query the API to see what's running on any instance â useful when self-hosting to verify your setup.
+
+ â See all engines
curl https://speech.ai.unturf.com/v1/models | jq .
@@ -221,46 +225,23 @@ make hydrate
The Five Engines
- ð§ Qwen3-TTS (tts-1-qwen) â Default
-
- - Quality: State-of-the-art voice cloning from 3-second samples
- - Languages: 10 languages (Chinese, English, Japanese, Korean, German, French, Russian, Portuguese, Spanish, Italian)
- - Voices: 42+ built-in cloned voices (aria, clara, elena, atlas, caleb, felix, and more)
- - Use Case: Default engine for all TTS â high quality with fast 97ms first-packet latency
- - Tech: 1.7B params, Apache 2.0 license, actively maintained by Alibaba
-
+ Each engine has a different personality. You don't need to think about model names â just pick a voice and we route it to the right engine. But if you're curious what's under the hood:
- ð Piper TTS (tts-1)
-
- - Speed: Real-time on CPU (fastest)
- - Voices: 100+ English voices via LibriTTS
- - Use Case: High-volume, low-latency synthesis
- - Tech: ONNX runtime, 22.05kHz output
-
+ ð§ Voice Cloning â Default Engine
+ Our default engine clones real human voices from 3-second audio samples. 42+ distinct speakers, each with their own timbre, cadence, and character. Speaks 10 languages natively â English, Chinese, Japanese, Korean, German, French, Russian, Portuguese, Spanish, and Italian. First audio arrives in under 100ms.
+ â Browse all 42+ voices
- ðïļ XTTS v2 (tts-1-hd)
-
- - Quality: Voice cloning from 6-second samples
- - Languages: 16 languages with auto-detection
- - Use Case: Custom voices, audiobooks, high-quality synthesis
- - Tech: Coqui TTS, requires ~4GB GPU VRAM
-
+ ð Fast Synthesis
+ When you need speed over everything else. Runs entirely on CPU, no GPU required. 100+ English voices. Ideal for high-volume batch jobs or real-time applications where latency matters more than naturalness.
- ⥠Silero TTS (tts-1-silero)
-
- - Speed: Real-time CPU inference
- - Voices: 148 voices across 5 languages (English, Russian, German, Spanish, French)
- - Use Case: CPU-only servers, multilingual applications
- - Tech: PyTorch, 48kHz output, actively maintained
-
+ ðïļ HD Voice Cloning
+ Clone any voice from a 6-second sample. Supports 16 languages with automatic detection. The highest fidelity option â great for audiobooks, character voices, or when you want to sound like yourself. Needs a GPU with ~4GB VRAM.
- ðŠķ Kokoro TTS (tts-1-kokoro)
-
- - Architecture: Lightweight decoder-only (82M params)
- - Voices: 34 intentionally OpenAI-themed voices (American/British English)
- - Use Case: Edge devices, low-resource environments
- - Tech: Apache 2.0 license, 24kHz output
-
+ ⥠Multilingual CPU
+ 148 voices across 5 languages (English, Russian, German, Spanish, French) â all running on CPU. The go-to choice when you need multilingual support without a GPU. Clean 48kHz output, actively maintained.
+
+ ðŠķ Lightweight
+ A tiny 82M parameter model that punches above its weight. 34 voices in American and British English. Built for edge devices, embedded systems, and anywhere resources are tight. Apache 2.0 licensed.
Get Involved