Kling Avatar 2.0 turns one character image and a recorded voice performance into a speaking-character video. The image establishes the character and scene, the audio supplies the spoken words and timing, and an optional prompt directs expressions and gestures. Choose Standard for a lower per-second rate or Pro when facial detail and expressive performance are priorities. Both tasks accept driving audio from 2 to 60 seconds, so you can prepare a portrait, record or generate speech, and animate the performance through the same input workflow.
Kling Avatar 2.0 is developed by Kuaishou. Vidgo provides access to the model. Model names and trademarks belong to their respective owners.Prepare one character image and a speech recording, then choose the version around your presentation goals and budget. Standard offers the lower per-second rate; Pro emphasizes a detailed character performance. Open either model card to supply the portrait, audio and optional directions for facial expression, posture and gestures.
Kling Avatar 2.0 Standard turns a single character portrait and driving audio into tightly synchronized talking-avatar video, with natural lip alignment, facial micro-expressions, and prompt-guided performance cues. It preserves facial identity and framing across the full generation cycle while delivering lively, coherent speaking posture and head motion.
Kling Avatar 2.0 Pro turns a character portrait and driving audio into high-fidelity talking-avatar video, with expressive gestures, object-interaction cues, and lifelike micro-movements. It preserves facial identity across longer speaking takes while refining eye motion, neck dynamics, and broadcast-ready presentation quality.
Both versions use the same image-and-audio workflow. Supply exactly one image to establish the character and framing, and one publicly accessible audio URL containing 2–60 seconds of speech. Add an optional prompt to describe the expression, movement and scene. Read each model’s existing introduction below, then open its task to prepare your materials and review the estimated cost.
| Model / task | Input and workflow |
|---|---|
| Kling Avatar 2.0 Standard | Kling Avatar 2.0 Standard turns a single character portrait and driving audio into tightly synchronized talking-avatar video, with natural lip alignment, facial micro-expressions, and prompt-guided performance cues. It preserves facial identity and framing across the full generation cycle while delivering lively, coherent speaking posture and head motion. |
| Kling Avatar 2.0 Pro | Kling Avatar 2.0 Pro turns a character portrait and driving audio into high-fidelity talking-avatar video, with expressive gestures, object-interaction cues, and lifelike micro-movements. It preserves facial identity across longer speaking takes while refining eye motion, neck dynamics, and broadcast-ready presentation quality. |
The portrait, speech and written directions each have a specific role. Use a clear character image for visual identity, an intelligible recording for words and timing, and a focused prompt for the performance you want to inspect in the generated video.
Upload one image that clearly shows the character and supply the speech recording. Kling Avatar 2.0 generates a speaking performance from these two inputs, with mouth movement following the audio. Review the greenhouse example to observe the relationship between the Mandarin speech, facial movement and the botanist’s small hand gesture.
Use the optional prompt to name the initial emotion, the change after a spoken pause and the closing expression. The actress example asks for vulnerability to soften into relief and a small smile. This gives you a concrete sequence to assess in the eyes, shoulders and face when reviewing the result.
Prepare a stylized character image and speech that suits the scene, then describe a simple gesture and expression. The recorded librarian example uses a three-dimensional character, Mandarin speech and a gentle calming gesture above a book. Review how the generated performance carries the original character design into a speaking scene.
Describe one purposeful movement and when the character should return to rest. In the astronomer example, the prompt asks for a single open-hand gesture toward the sky and a small closing smile while the supplied English audio plays. Use a similarly explicit action to guide a presentation, explanation or character monologue.
Use clear speech from 2 to 60 seconds in a publicly accessible audio file of up to 5 MiB. Choose Standard or Pro and review the rate before submission. The service detects the audio duration and rounds it up to a whole second for billing, so the voice performance determines both the shot’s timing and its billable duration.
The Standard example uses a botanist in a greenhouse, Mandarin speech and a prompt requesting a small gesture toward new leaves followed by a warm smile. The Pro example uses an actress and a quiet personal confession, with directions for a change from vulnerability to relief. Play the input audio and output video to inspect mouth movement, pauses, expressions and gestures. Each example has its own source materials and creative brief, giving you two recorded uses of the shared workflow.
The botanist speaks gently in Mandarin, making one small open-hand gesture toward the new leaves. Her patient expression becomes a warm smile after the pause. Accurate lip sync, natural blinking, stable identity and hands. Locked camera, no cuts.
The actress delivers a quiet personal confession. Her initially vulnerable expression softens into relief and a small sincere smile after the pause. One subtle exhale, relaxed shoulders, hands remain natural. Precise lip sync and nuanced eyes, stable identity. Locked camera, no cuts.
Start with the character image, the words to be spoken and the intended emotional tone. Record or generate the speech, describe a short performance, and use the resulting video in the project that needs a visible speaker.
Prepare a presenter portrait and a concise narration about your subject. Supply the audio, describe an attentive expression and a small explanatory gesture, then generate a speaking clip for a lesson introduction, a product explanation or an informational presentation.
Prepare a character image, record the dialogue and identify the emotional change within the scene. Describe the initial expression, any pause and the final reaction. Review the video as a performance study for a narrative concept or a short character introduction.
Choose a drawing or three-dimensional character image and record a passage in the character’s voice. Describe a gentle gesture and facial expression that suit the story. Use the speaking result to explore a story introduction, an illustrated narration or an animated character scene.
Keep the chosen character image and prepare separate speech recordings for each language. Generate a video for each recording with directions appropriate to the spoken message. Review pronunciation timing, mouth movement and expressions before using the clips in a multilingual presentation.
Record a line with deliberate pacing and pauses, then describe how the face and posture should change across the performance. Generate a clip and inspect the eyes, mouth, shoulders and timing. Use the result to discuss an intended emotional direction with a creative team.
Prepare reusable character images and a process for recording or generating speech. Send the selected version, image, audio and prompt through the API, save the task ID and retrieve the completed video. This connects character preparation, voice production and video delivery in your application.
Both versions are billed by the detected duration of the driving audio, rounded up to a whole second. Multiply that billable duration by the selected version’s current per-second rate. The dynamic table shows the starting rate and shared input limits; open the task page with your actual audio to review the estimate. File metadata can include a fractional second beyond the rounded duration shown by a player, so use the service’s detected duration when calculating usage.
| Model | Provider | Version | Character images | Driving audio duration | Starting rate |
|---|---|---|---|---|---|
| Kling Avatar 2.0 Standard | Kling | Standard | 1 | 2–60 s | $0.035 / second |
| Kling Avatar 2.0 Pro | Kling | Pro | 1 | 2–60 s | $0.07 / second |
Prepare the portrait and speech, review the performance, then use the same inputs in your API integration.
Open a task, supply one character image and an accessible 2–60-second audio file, and describe the intended expression or gesture. Review the estimate, submit the generation and play the result to inspect mouth movement, pauses and the character’s appearance.
Vidgo brings the two Avatar versions, online input controls, current rates and task documentation into one workflow. Test a portrait with its speech, review the result and carry the selected model and materials into your application.
Review Standard and Pro together, including their existing model introductions and current per-second rates. Open the appropriate model card to test the performance that matters to your project.
Supply your character image and audio on the task page, then add directions for expression and gestures. Play the generated video and review the mouth movement, pauses and character appearance.
Check the detected audio duration and the selected model’s rate before submitting. This helps you plan a testing budget around the speech recordings your application will actually process.
Use the matching task documentation for authentication, model identifiers, media fields and optional prompts. Prepare a publicly accessible audio URL and an accepted image source for each request.
Retain the returned task ID and query its status until the video completes. Provide a callback URL when your application should receive a completion notification and the resulting file information.
Use generation history to review previous requests and completed outputs from online testing. Keep the chosen portrait, speech and performance directions together when preparing another version of the scene.
Kling Avatar 2.0 is Kuaishou’s audio-driven character-video model. It combines one character image with a speech recording to generate a speaking performance, and accepts an optional prompt for expressions, gestures and scene direction. Vidgo provides Standard and Pro tasks with the same input workflow and their own per-second rates.
Read Vidgo API guides and platform comparisons to understand task integration, model selection and usage costs when building a character-video workflow.
