HeyGen Lip Sync (Speed)

使用此模型的任务

输入合同

必需输入

audio, video

可选输入

disable_music_track, enable_dynamic_duration, enable_speech_enhancement

输出

video (uri)

提交方式

Replicate 官方模型端点（始终为当前版本）

供应商计费

合同备注

Re-locked 2026-10-06 (batch E): latest version 8eabe9f38713 carries a field-identical schema vs the A0 snapshot; schemaVersion updated in lockstep. enable_dynamic_duration is PINNED OFF server-side: with it on, the output follows the audio length — billing seconds the upload layer cannot probe — and the provider enforces a ±15% audio/video duration preflight that rejects mismatched pairs outright (probe run 2026-10-06: 3.00 s video + 4.73 s audio failed preflight). With it off, output = source video length and longer audio is trimmed (MEASURED 2026-10-06: 3 s video + 4.733 s audio → 3.000 s / 1920x1080 output). "Fast" mode of the lip-sync task; not an automatic retry of the precision model. Enabled 2026-10-06: batch E deployed-app E2E green (batch-e-e2e report).

状态

来源

• 供应商 schema
• 供应商 README