New Audio Models in Pixraft: Qwen3 TTS, Seed Audio, and MiniMax Music 3.0

New audio options are now available

The latest Pixraft catalog refresh adds Qwen3 TTS, ByteDance Seed Audio 1.0, and MiniMax Music 3.0 to the supported audio model set. They solve different problems: spoken narration, guided speech and audio experiments, and complete music or instrumental beds.

The goal is not to suggest that one model replaces a voice actor, sound designer, composer, or mix engineer. These models give you more starting points inside the same creative workspace.

Qwen3 TTS: narration with voice and style controls

WaveSpeed documents Qwen3 TTS with 11-language support, 9 preset voices, and optional style instructions. In Pixraft’s Audio Studio, the model is suited to turning a script into narration, a draft voice-over, an accessibility track, or spoken dialogue for a video concept.

Use short test passages first. A style instruction such as “calm, clear, and slightly slower for an instructional video” communicates delivery intent, but it does not guarantee a specific performance or pronunciation in every sentence.

Seed Audio 1.0: guided speech and audio experiments

Seed Audio accepts a prompt and can use either reference audio or a reference image, along with controls for format, sample rate, speed, volume, and pitch. The model page says the audio and image guidance paths should not be combined, so follow the fields shown in Pixraft’s model form.

This can help with narration prototypes, dialogue concepts, image-guided sound ideas, and quick audio variations. Treat reference audio carefully: only upload material you have permission to use, and do not imitate a real person’s voice without consent and a lawful basis.

MiniMax Music 3.0: songs and instrumental beds

MiniMax Music 3.0 generates music from a style prompt and lyrics, with structure tags such as [Verse], [Chorus], and [Bridge]. WaveSpeed also documents an instrumental-only mode and configurable bitrate and sample rate settings.

That makes it useful for demos, background beds, rough soundtrack directions, podcast intros, and short-film music sketches. It can reduce the distance between a written mood and an audible idea, while a final release may still require arrangement, editing, mixing, mastering, and a rights review.

How Pixraft helps you use them

1. Open Audio Studio and choose the model that matches the output: speech, guided audio, or music.
2. Test a short input before spending time on a full script or song.
3. Set the language, voice, style, output format, or music controls that the selected model exposes.
4. Review Pixraft’s live credit estimate before generating and keep the input version with any result you want to reproduce.
5. Combine the output with Video Studio, Lip Sync Studio, or Clipping Studio when the audio belongs to a larger production.
6. Download, edit, and review the result in the toolchain appropriate to your delivery format.

What Pixraft does not promise

Audio generation can contain pronunciation errors, unwanted artifacts, timing problems, or musical choices that do not fit the brief. A generated voice is not automatically cleared for every commercial, likeness, or identity use. A generated song is not automatically free of third-party rights questions.

Check the provider license, your input permissions, and the final output before publishing. Model schemas, availability, and credit estimates can change as providers update their services; use the live Pixraft form for the current configuration.

Why this matters for Pixraft users

The new arrivals make Pixraft more useful when a project crosses media types. You can develop a still, animate it, create narration or music, synchronize dialogue, and cut variants without moving every experiment to a separate model account. Pixraft does not guarantee the creative result, but it gives you a connected place to test the model that fits the next production decision.