ComfyUI Course 1-10

Останні два відео згенеровані у workflow LTX-2 Image to Video, в промті прописаний текст голосу жінки і описані рухи жінки. Голос не є чітким, роботизований і якість зображення у відео гірше ніж на Wan2.2, але генерує швидше.
Іноді жінка не спілкується голосом. В LTX раніше були такі недоліки з якістю відео.
В LTX-2 з'явилась можливість завантажити зображення і аудіо.

Це відео на LTX-2 Image to Video with Custom Audio, вийшло досить вдало. Відео 704x1280 6 секунд по зображенню і аудіо генерувалося у хмарі 5 хвилин (24 GB Vram), на PC буде набагато довше генерувати якщо відеокарта слабша.
При генерації цього воркфлоу на PC у мене була помилка.

1. Qwen3-TTS CustomVoice    Скріншот воркфлоу
We're using the Ryan preset for stable narration, then saving straight to MP3. Pretty clean, right?

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
What's up, I'm Pixaroma. In this workflow we design a custom voice. Instead of picking a preset speaker, you describe it.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
I am sorry, that was funnier than it should be.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
I thought I was ready, but I was wrong.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
You promised. You lied. You mother Beeeep...

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Do you hear that, behind us?

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
In a world where nothing makes sense, one choice changes everything.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Team, eyes up. We move in ten seconds.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Muhahaha, the plan is flawless, mostly.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Hi friend, want to try something yummy today?

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
I do not like it. I do not trust it.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Shh, listen, the stars are talking tonight.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Okay okay okay, we run, then we hide, then snacks!

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Come closer. I want to tell you something.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Good morning, you are listening to Starwave, stay with us.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Fresh today, priced right, come in and save.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Let your shoulders drop, and let your thoughts pass by.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
Listen to that. Soft. Clean. Perfect.

2. Qwen3-TTS VoiceDesign    Скріншот воркфлоу
I am having so much fun with this workflow. Are you ready for the next one?

3. Qwen3-TTS VoiceClone    Скріншот воркфлоу
Reference Audio Text (Must Match Voice Sample Audio):
Hey, I'm AI generated. Be honest, do you like my voice and does it feel real?
Speech Text:
Welcome to Pixaroma Radio. This is Alison speaking. Today we're testing a cloned female voice inside ComfyUI. Pretty cool, right?

3. Qwen3-TTS VoiceClone    Скріншот воркфлоу
Reference Audio Text (Must Match Voice Sample Audio):
I am having so much fun with this workflow. Are you ready for the next one?
Speech Text:
Welcome to Pixaroma Radio. This is Alison speaking. Today we're testing a cloned female voice inside ComfyUI. Pretty cool, right?

4. Qwen3-TTS VoiceClone + SaveVoice    Скріншот воркфлоу
Reference Audio Text (Must Match Voice Sample Audio):
Hey, I'm Al generated. Be honest, do you like my voice and does it feel real?
Speech Text:
Welcome to Pixaroma Radio. This is Alison speaking. Today we're testing a cloned female voice inside ComfyUI. Pretty cool, right?

5. Qwen3-TTS VoiceDesign + SaveVoice    Скріншот воркфлоу
What's up, I'm Pixaroma. In this workflow we design a custom voice. Instead of picking a preset speaker, you describe it.

6. Qwen3-TTS VoiceClone + Load A Previous Saved Voice    Скріншот воркфлоу
Welcome to Pixaroma Radio. Today we're testing a cloned voice inside ComfyUI. Pretty cool, right?

6. Qwen3-TTS VoiceClone + Load A Previous Saved Voice    Скріншот воркфлоу
Welcome to Pixaroma Radio. Today we're testing a cloned voice inside ComfyUI. Pretty cool, right?

7. Qwen3-TTS VoiceClone + Load Saved Multi Voice + Dialogue    Скріншот воркфлоу
man: Hey, you there?
woman: Yeah, I'm here, what are we testing?
man: A clean TTS workflow in ComfyUI
woman: Nice, voice design or cloned voice?
man: Cloned voice for consistency
woman: Love it!

8. Qwen3-TTS VoiceDesign + Dialogue    Скріншот воркфлоу
man: Hey, you there?
woman: Yeah, I'm here, what are we testing?
man: A clean TTS workflow in ComfyUI
woman: Nice, voice design or cloned voice?
man: Cloned voice for consistency
woman: Love it!

1. Fish S2 Voice Clone TTS    Скріншот воркфлоу
[happy] Hey, this is Pixaroma, and [laughing] okay, this is already making me way too happy. [giggle] There is just something so fun and exciting about all of this. [Inhale] So come on, let us enjoy every second together. [crowd laughing] [background laughter]

2. Fish S2 TTS    Скріншот воркфлоу
[A cheerful female voice with a joking, playful tone] [happy] I clicked one button, everything broke, and for a second .. [pause]
I just stared at the screen! [pause], [laughing] [laughing] Honestly, it was kind of impressive.

3. Fish S2 Multi-Speaker TTS    Скріншот воркфлоу
[speaker_1]: [excited] So, what are we doing today?
[speaker_2]: [professional broadcast tone] Something simple, but really useful.
[speaker_1]: [happy] Nice, I like the sound of that
[speaker_2]: [inhale]Then let us get started [laughing][chuckle].