ComfyUI-QwenTTS/Update.md

update v1.1.4

• new workflow work with new released ComfyUI-QwenAR Workflow

update v1.1.3

• Dependence update and bug fixed

Update v1.1.0

Highlights

• Voice Clone nodes now support VOICE inputs from the Voices Library for stable reuse.
• New Tools bundle: Create Voice, Load Voice, Whisper STT, and Voice Instruct presets (EN + CN).
• Advanced nodes now expose attention selection (auto / sage_attn / flash_attn / sdpa / eager).
• README expanded with ComfyUI extra_model_paths.yaml guidance for custom model locations.
• Audio Duration node rewritten: cleaner logic, seconds-based outputs, optional frame calculation.
• 5 new sample workflows (v1.1.0):
  • Clone_Voice.json — Voice Clone demo workflow (2 variants).
  • Create_your_voice.json — Create and save your own voice profile.
  • QwenTTS_Nodes.json — Overview of all custom nodes in this repo.
  • QwenTTS_sample_workflow.json — Sample nodes for Voice Clone, Voice Design, and Custom Voice.
  • Voice_design.json — Advanced Voice Design + preset voice instruct nodes (EN/ZH).

New / Updated Nodes

• Create Voice (QwenTTS)
  • Build and save voice prompts to .pt in ComfyUI/output/qwen3-tts_voices by default.
• Load Voice (QwenTTS)
  • Load saved voices or use a custom path and output VOICE.
• Whisper STT (QwenTTS)
  • Transcribe AUDIO to text with multiple model sizes.
• Voice Clone (QwenTTS) / Voice Clone (QwenTTS) Advanced
  • Added optional voice input; reference_audio is only required if no voice is provided.
• Voice Instruct (QwenTTS)
  • English preset builder from voice_instruct.json.
• 声音风格指引 (QwenTTS)
  • Chinese preset builder from voice_instruct_zh.json.

Audio Duration

• Outputs: duration_int (seconds), duration_float (seconds), frames, audio_path.
• Optional fps input enables frame calculation.