slim-orpheus-3b-JAPANESE-ft

Slim-Orpheus 3B Japanese

Pruned original weights down from 28 -> 16 layers (43% reduction) to speed up inference and reduce memory requirements. Trained in Japanese on 14 voices.

Voices

Below are sample outputs for each voice with quality indicators:

• ⭐⭐⭐ Good quality
• ⭐⭐ Okay quality
• ⭐ Poor quality
• ⚠️ Unstable

Lyney ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Lyney.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Cyno ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Cyno.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Tighnari ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Tighnari.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Kaeya ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Kaeya.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Neuvillette ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Neuvillette.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Kaveh ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Kaveh.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Dehya ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Dehya.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Yae Miko ⭐⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Yae_Miko.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Layla ⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Layla.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Yoimiya ⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Yoimiya.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Alhaitham ⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Alhaitham.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Zhongli ⭐⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Zhongli.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Furina ⭐

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Furina.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Arataki Itto ⚠️

<audio controls> <source src="https://huggingface.co/Gapeleon/slim-orpheus-3b-JAPANESE-ft/resolve/main/samples/Arataki_Itto.wav" type="audio/wav"> Your browser does not support the audio element. </audio>

Limitations

• Japanese Only: This model was trained specifically for Japanese language and cannot speak English or other languages
• No Emote Support: Not trained on emotes/emotional cues like <happy>, <sad>, etc. that were available in the original model
• Reduced Parameter Count: While offering faster inference, the reduction from 28 to 16 layers may impact some of the nuanced capabilities of the original Orpheus model
• Voice Quality Varies: As noted in the voice quality ratings, some voices perform better than others

---

Orpheus-TTS Model Details:

Code is available on GitHub: CanopyAI/Orpheus-TTS

Orpheus TTS is a state-of-the-art, Llama-based Speech-LLM designed for high-quality, empathetic text-to-speech generation. This model has been finetuned to deliver human-level speech synthesis, achieving exceptional clarity, expressiveness, and real-time streaming performances.

Model Details

Model Capabilities

• Human-Like Speech: Natural intonation, emotion, and rhythm that is superior to SOTA closed source models
• Low Latency: ~200ms streaming latency for realtime applications, reducible to ~100ms with input streaming

Original Model Sources

• GitHub Repo: https://github.com/canopyai/Orpheus-TTS

Usage

Check out the Orpheus Colab: (link to Colab) or GitHub (link to GitHub) on how to run easy inference on our finetuned models.

Model Misuse

Do not use our models for impersonation without consent, misinformation or deception (including fake news or fraudulent calls), or any illegal or harmful activity. By using this model, you agree to follow all applicable laws and ethical guidelines. We disclaim responsibility for any use.