翻译字幕阶段出错 [兼容AI/本地模型] 无返回翻译结果
self=LocalLLM(uuid='f8a3e0ad5d', proxy_str=None, last_down_time=0, translate_type=9, text_list=[SrtItem(text='ご視', start_time=6740, end_time=8280, startraw='00:00:06,740', endraw='00:00:08,280', line=1, time='00:00:06,740 --> 00:00:08,280', spk='', filename=''), SrtItem(text='聴ありがとうございました', start_time=8280, end_time=29860, startraw='00:00:08,280', endraw='00:00:29,860', line=2, time='00:00:08,280 --> 00:00:29,860', spk='', filename=''), SrtItem(text='おはようございます おはよう\nお食事お待ちしました ありがとう\nお飲み物お待ちします', start_time=40420, end_time=48520, startraw='00:00:40,420', endraw='00:00:48,520', line=3, time='00:00:40,420 --> 00:00:48,520', spk='', filename=''), SrtItem(text='失礼します ありがとう', start_time=69080, end_time=72400, startraw='00:01:09,080', endraw='00:01:12,400', line=4, time='00:01:09,080 --> 00:01:12,400', spk='', filename=''), SrtItem(text='レイナさん綺麗', start_time=80000, end_time=83660, startraw='00:01:20,000', endraw='00:01:23,660', line=5, time='00:01:20,000 --> 00:
......
, and complex connectors. Prefer the shortest, punchiest spoken phrasing.\n\n3. NATURAL SPOKEN REGISTER:\n - Write for the ear, not the eye. Use everyday conversational vocabulary heard in dubbed cinema. Avoid literary syntax or literal translationese.\n - Target Language Rules for Simplified Chinese:\nUse natural colloquial Mandarin (普通话口语). Drop redundant subject pronouns (我, 你, 他) when clear from context. Strip translationese (avoid \'关于...\', \'进行了...\'). Target a dubbing pace of 3-4 characters per second. For split fragments, use natural spoken sentence-final particles (吧, 呢, 嘛) or spoken inversion (倒装) to keep each block self-contained without moving words.\n\n
\n\n# FEW-SHOT DEMONSTRATION OF STRICT 1-TO-1 MAPPING\n\nSource Input (Split sentence + Short reaction):\n1\n00:00:01,000 --> 00:00:02,300\nI think we should definitely\n\n2\n00:00:02,350 --> 00:00:03,800\nleave this place right now.\n\n3\n00:00:04,000 --> 00:00:04,800\nYeah.\n\nCORRECT OUTPUT (Count=3, Timestamps exact, Zero-shifted, Ellipsis bridged):\n\n1\n00:00:01,000 --> 00:00:02,300\n[Translated fragment of "I think we should definitely"...]\n\n2\n00:00:02,350 --> 00:00:03,800\n[...Translated fragment of "leave this place right now."]\n\n3\n00:00:04,000 --> 00:00:04,800\n[Translated "Yeah."]\n\n\n
\n\n\n\n\n# FINAL OUTPUT CONSTRAINTS\n- Output ONLY the valid SRT text wrapped inside
and .\n- Do NOT output markdown code blocks (NO or srt).\n- Do NOT add any conversational comments, greetings, or explanations.\n- Execute block-by-block. Ensure Block 1 translates Input 1, Block 2 translates Input 2... up to the final block.\n\n\n{batch_input}\n', api_key='1234', temperature=0.1, max_tokens=8192, reasoning_effort=None, extra_body=None)[TranslateSrt]翻译字幕: OutputFormat=单语字幕
[TaskCfgSTS]
原始输入文件名: D:/RUANJIAN/win-pyvideotrans-v4.14/output/recogn/MIDA-241.H265.srt,
输出结果保存到文件夹: D:/RUANJIAN/win-pyvideotrans-v4.14/output/recogn
翻译渠道:兼容AI/本地模型,原始语言:ja,目标语言:zh-cn, 已选 发送完整字幕