ResearchBenchmarksTalking-head Generation on Talking-head rendering CPU-onlyTalking-head Generation on Talking-head rendering CPU-onlyFollow26.67Latency (ms)VedicTHG-10.5548240.7126491.98743.2474Feb 9, 2026Evaluation resultsMethodMethodLinksLatency (ms)FPSPeak CPU Utilization (%)VedicTHGHardware=16-core CPU,...Hardware=16-core CPU, Pipeline Stage=Render-only (renderer + compositing)2026.0226.6737.529.25Wav2LipHardware=16-core CPU,...Hardware=16-core CPU, Pipeline Stage=Render-only (renderer + compositing)2026.02957.291.04811