Kling AI Updates Video Models With Timbre Control and Enhanced Motion Capabilities

Kling AI introduced a series of updates to its generative video platform between December 15 and December 17, deploying new features for audio consistency, complex motion rendering, and flexible narrative timing. The releases specifically target the Kling Video 2.6 and Video O1 models, addressing technical challenges related to character dubbing and extended action sequences.

Audio Consistency and Timbre Control

The Kling Video 2.6 model has integrated a "Timbre Control" feature designed to maintain vocal consistency across generated content. Building on the platform's existing synchronized audio-video generation, this update allows for the stabilization of specific voice tones throughout a clip.

According to the release notes, the system enables users to bind distinct vocal characteristics to specific figures within a scene using prompt instructions formatted as "Character @TimbreName." This mechanism is intended to ensure that a character's voice remains uniform from start to finish, supporting applications such as virtual avatars, product demonstrations, and multi-character dialogue.

Extended Motion and Action Sequences

Kling AI also upgraded the "Motion Control" capabilities within the 2.6 model, focusing on the generation of complex physical movements. The system can now produce continuous shots lasting up to 30 seconds without the need to split footage.

The update reportedly improves the model's responsiveness to high-difficulty actions, including martial arts sparring, sports competitions, and dance routines. Technical improvements cite better full-body synchronization and more refined rendering of limb and hand details during these extended sequences.

Video O1 Model Adjustments

The Video O1 model received parallel updates aimed at increasing production flexibility. A new 720p resolution mode has been added, which retains the core functional capabilities of the standard 1080p model while facilitating lightweight content creation.

Furthermore, the update modifies the handling of first and last frames. Users can now define narrative durations between 3 and 10 seconds for these segments, replacing previous fixed limits with variable timing options for the start and conclusion of generated videos.