Hongguo Short Drama Transcript API | JSON & Hooks

🎙️ 红果短剧台词转写 API / Hongguo Short Drama Transcript API

红果短剧台词转写 API 将 红果短剧 Hongguo Duanju 的单集对白转成 JSON，返回完整识别文本、时间轴、说话人编号，以及表示第一句识别台词的 hook（钩子）。每次支持最多 10 部剧或 1,000 个单集。

Hongguo Short Drama Transcript API converts Chinese short-drama dialogue into JSON transcripts with timestamps, speaker labels, and opening-sentence hook. Process up to 10 series or 1,000 episodes per run for dialogue search, opening-line research, and content analysis.

🔗 配套红果工具 / Related Hongguo APIs

• 📚 剧集详情与单集 ID / Hongguo Series Details & Episodes API — 获取完整目录，选择需要转写的剧集。 Get episode directories and IDs for transcription.
• 🔎 短剧搜索 / Hongguo Short Drama Search API — 按剧名或关键词查找剧集 ID。 Find series IDs by title or keyword.
• 🔥 推荐短剧 / Hongguo Recommendation Feed API — 发现推荐内容后分析对白。 Discover recommended dramas for dialogue research.
• 📅 短剧上新日历 / Hongguo Release Calendar API — 按日期发现新剧和待上线内容。 Discover releases and upcoming titles by date.
• 💡 搜索关键词建议 / Hongguo Keyword Suggestions API — 扩展搜索词，再查找对应剧集。 Explore search suggestions before finding series.

推荐使用剧集 ID（seriesIds）。 Actor 会自动获取剧名和简介作为识别上下文，可能改善人物名称与剧情术语的识别。单集 ID（videoIds）也可使用，但不附加这些剧集信息。上下文不代表准确率保证。

Use series IDs (seriesIds) when available. The Actor includes the series title and introduction as recognition context, which may improve character-name and story-specific terminology recognition. Episode IDs (videoIds) work without additional series context. Accuracy improvement is not guaranteed.

✨ 可以用来做什么？ / What can you do with Hongguo transcripts?

• 台词检索： 为已获授权的短剧建立可搜索的对白资料。 Dialogue search: build searchable transcripts for authorized short-drama research.
• 开场分析： 比较不同剧集的首句台词和出现时间。 Opening-line analysis: compare the first recognized sentence and its timing across episodes.
• 内容研究： 结合剧集元数据分析主题、术语和对白。 Content research: join episode transcripts with series metadata to study themes and dialogue.
• 自动化工作流： 从 Apify Console 操作，或通过 API 读取结构化结果。 Automation: run from Apify Console or consume structured results through the API.

说话人编号用于区分同一集里的声音，不代表人物姓名，也不保证不同集的同一编号属于同一个人。

Speaker labels distinguish voices within one episode. They do not identify character names or match the same person across episodes.

🚀 如何转写红果短剧？ / How do I transcribe Hongguo episodes?

1. 从上方的搜索、推荐或日历工具获取剧集 ID；也可用剧集详情工具获取单集 ID。 Find a series ID using Search, Recommendations, or Calendar, or obtain episode IDs from Series Details.
2. 在 Input 中填写 seriesIds 或 videoIds，一次只使用一种。 Enter either seriesIds or videoIds in Input; use one non-empty list per run.
3. 在 Apify 运行设置中设置最高花费并启动 Actor。 Set your maximum run spend in Apify run settings and start the Actor.
4. 在 Dataset 查看每集 JSON，词级时间戳可通过 wordsJsonUrl 获取。 Read one JSON row per unique episode and follow wordsJsonUrl for detailed word timestamps when available.

剧集模式最多接受 10 个唯一剧集 ID，展开后最多 1,000 个唯一单集。单集模式最多接受 1,000 个唯一视频 ID。超出上限的请求会报错，需拆分运行；不会静默截断。

Series mode accepts up to 10 unique series IDs and up to 1,000 unique episodes after expansion. Episode mode accepts up to 1,000 unique video IDs. Over-limit requests are rejected and must be split into separate runs; they are not silently truncated.

📥 输入示例 / Input examples

推荐：使用剧集 ID / Recommended: series IDs

运行表单预填了下方剧集 ID，单集列表保持为空。若改用 videoIds，请先清空 seriesIds。

The run form prefills the series ID below and leaves episode IDs empty. To use videoIds, clear seriesIds first.

{
  "seriesIds": [
    "7678236482489830424"
  ]
}

此 ID 在 2026-11-05 的实测中对应《垂竿向晚》。内容可用性可能变化，请通过剧集详情工具确认最新目录。

This series ID resolved to 垂竿向晚 during testing on November 5, 2026. Availability can change; use Series Details to check the current episode directory.

仅转写指定单集 / Transcribe a specific episode

{
  "videoIds": [
    "7678238971456932889"
  ]
}

这是上述剧集在该次测试中返回的首集 ID。两段 JSON 是独立请求，不要合并为同时非空的两个列表。

This was the first episode ID returned for that series in the same test. These JSON examples are separate requests; do not combine them into two non-empty lists.

参数 / Field	说明 / Description
seriesIds	最多 10 个唯一剧集 ID，自动附加剧名和简介。 Up to 10 unique series IDs, with automatic title and introduction context.
videoIds	最多 1,000 个唯一单集 ID，不附加剧集上下文。 Up to 1,000 unique episode IDs, without additional series context.

ID 必须是数字字符串。重复 ID 会先去重，再检查数量限制。输入只接受这两个字段，不提供手动上下文、热词、输出格式或缓存策略参数。完整约束见 Input 页。

IDs must be numeric strings. Duplicates are removed before limits are checked. Only these two input fields are accepted; there are no manual context, hotword, output-format, or cache-policy settings. See the Input tab for full validation rules.

📤 JSON 转写结果 / JSON transcript output

Dataset 每个唯一单集一行。以下为虚构的结构示例，不是实际识别结果或准确率评测。

The Dataset contains one row per unique episode. The fictional example below illustrates the structure; it is not a measured transcription result or an accuracy benchmark.

{
  "seriesId": "1234567890123456780",
  "videoId": "1234567890123456789",
  "episodeNumber": 1,
  "title": "示例短剧",
  "sourceSeriesIds": [
    "1234567890123456780"
  ],
  "durationSeconds": 120.0,
  "language": "zh",
  "transcript": "你怎么来了？我来找你。",
  "hook": "你怎么来了？",
  "hookStartSeconds": 1.2,
  "hookEndSeconds": 2.8,
  "segments": [
    {
      "start": 1.2,
      "end": 2.8,
      "text": "你怎么来了？",
      "speakerId": "speaker_0"
    },
    {
      "start": 3.4,
      "end": 4.9,
      "text": "我来找你。",
      "speakerId": "speaker_1"
    }
  ],
  "wordsJsonUrl": null,
  "status": "succeeded",
  "errorCode": null,
  "generatedAt": "2026-11-05T00:00:00Z"
}

字段 / Fields	含义 / Meaning
seriesId, videoId, episodeNumber, title, sourceSeriesIds	剧集与单集关联信息，未知值为 null 或空列表。 Series and episode associations; unavailable values are null or an empty list.
durationSeconds, language, transcript	完整音频时长、可用时的语言标识和识别台词。 Full audio duration, language when supplied, and recognized dialogue.
hook, hookStartSeconds, hookEndSeconds	首句识别文本及可靠的时间边界，单位为秒。 First recognized sentence and reliable timing boundaries in seconds.
segments	句段的 start、end、text、speakerId，保留真实重叠。 Segment timestamps, text, and speaker IDs; overlapping speech timing is preserved.
wordsJsonUrl	当前运行的词级 JSON 链接，不可用时为 null。 This run's word-level JSON link, or null when unavailable.
status, errorCode	处理结果和错误原因。 Processing outcome and error reason.
generatedAt	转写文本最初生成的时间，使用缓存时可能早于本次运行。 When the transcript was originally generated; it may predate this run when a cached result is reused.

本 Actor 交付 JSON，不生成 SRT/VTT 字幕或视频下载文件。Apify 通用 Dataset 导出界面可能提供其他格式，这不代表 Actor 支持输出格式选择。

This Actor delivers JSON and does not create SRT/VTT subtitles or downloadable video files. Apify's generic Dataset export interface may offer other formats; this does not add output-format options to the Actor.

💬 hook（钩子）抽取 / Opening hook extraction

批量抽取红果短剧开场钩子，让引流素材研究从首句台词开始。 输入剧集 ID，即可随完整转写获取每集的 hook（钩子）和可用时间边界，快速整理开场台词库、比较悬念与冲突表达，为短视频引流文案、剪辑选段和素材测试提供参考。结合上方的搜索、推荐和剧集详情工具，可串联找剧、选集与钩子抽取流程；钩子抽取已包含在单集价格中。

Extract opening hooks from Hongguo short dramas in bulk. Submit series IDs to collect each episode’s opening hook and available timing alongside its full transcript. Build an opening-line library, compare suspense and conflict, and inform promotional copy, clip selection, and creative testing. Use the Search, Recommendations, and Series Details APIs above to move from discovery to hook extraction, included in the episode price.

hook 是音频时间顺序中第一句非空的识别台词，类型为字符串或 null。它不是生成的营销文案、剧情摘要或固定的前三秒片段。开头的静音不会占用首句位置。

hook is the first non-empty recognized sentence in audio order, returned as a string or null. It is not generated marketing copy, a plot summary, or a fixed three-second excerpt. Opening silence does not create an empty hook.

一个识别段包含多句话时，按句末标点取首句；没有句末标点时保留首个识别片段。若无法可靠确定切分后的结束时间，hookEndSeconds 为 null。失败或无语音时，hook 为 null。

When a segment contains several sentences, the Actor takes the first sentence ending in punctuation. Without sentence-ending punctuation, it returns the first recognized fragment. If the exact end cannot be aligned reliably, hookEndSeconds is null. Failed or speech-free episodes have hook: null.

💳 价格、配额与重复运行 / Pricing, quotas, and repeat runs

以下为当前配置的事件价格；启动前以 Apify Pricing 页显示的有效价格为准。

The event prices below match the current configuration. Check the active prices on the Apify Pricing tab before starting a run.

收费项目 / Event	价格 / Price
付费 Apify 账户的成功单集，含前 180 秒 / Successful episode on paid Apify plans, including 180 seconds	$14.99 / 1,000 集 · episodes ($0.01499 / 集 · episode)
Apify Free 账户的成功单集，含前 180 秒 / Successful episode on the Apify Free plan, including 180 seconds	$14.99 / 1,000 集 · episodes ($0.01499 / 集 · episode)
超过 180 秒后的每个开始分钟 / Each started minute beyond 180 seconds	$0.006 / 分钟 · minute
失败、无语音、未完成或预算跳过 / Failed, speech-free, pending, or budget-skipped episode	不收成功结果事件费 / No successful-result event charge

上述价格仅为 Actor 事件费用；Apify 平台运行、存储、代理等使用成本由用户另行承担，免费与付费账户均适用。

These prices cover Actor event charges only. Users also pay Apify platform usage costs, including compute, storage, and proxies, on both Free and paid plans.

JSON 和 hook 已包含在单集价格中。例如，付费账户转写一集 4 分 20 秒的内容，事件费用为 $0.01499 + 2 × $0.006 = $0.02699。整剧输入也是逐集计算。

JSON and hook are included in the episode price. A 4-minute-20-second episode on a paid plan costs $0.01499 + 2 × $0.006 = $0.02699 in event charges. Series inputs are also priced per episode.

Apify Free 账户每个 UTC 自然月最多获得 10 个成功单集结果。 这是使用上限，不是 10 次免费转写；上述事件价格仍适用。缓存命中和新运行中的重复购买都计入配额。

Apify Free-plan accounts are limited to 10 successful episode results per UTC calendar month. This is a usage cap, not 10 free transcripts. The event prices still apply, and cache hits and successful repeat runs count toward the cap.

同一运行内的重复视频 ID 只输出一次、收取一组成功结果费用。每次新运行正常收费，包括已购买的单集和缓存结果。 单次输入上限、月度配额、运行最高花费是三个独立限制。

Duplicate video IDs in one uninterrupted run produce one result and one set of successful-result charges. Every new run is billed normally, including previously purchased episodes and cache hits. Input limits, monthly quotas, and the maximum run spend are separate constraints.

⚙️ 准确性、缓存与处理边界 / Accuracy, caching, and processing limits

剧集模式会把标题和简介整理为最多 400 个字符的上下文；单集模式仅依据音频进行识别。背景音乐、多人重叠、方言和音质都可能影响结果，重要内容建议人工复核。

Series mode supplies up to 400 characters of title and introduction context. Episode mode recognizes the audio without additional metadata context. Music, overlapping voices, dialects, and audio quality can affect results; review transcripts when accuracy is important.

视频、来源版本和识别配置匹配时可复用缓存。缓存结果保留原始生成时间；有剧集上下文与无上下文的结果分别匹配，缓存不会自动降低单集价格。

A cached result may be reused when the video, source version, and recognition configuration match. It retains its original generation time. Results with and without series context are matched separately, and cache reuse does not reduce the episode price.

单集最长 30 分钟，源文件最大 512 MiB。运行预算在处理前预留；时长未知时采用保守预留，因此较低预算可能跳过一个实际较短的未缓存单集。

Each episode is limited to 30 minutes and a 512 MiB source file. Budget is reserved before processing, conservatively when duration is unknown, so a low budget may skip an uncached episode that is actually shorter.

当前不支持中断续跑或自动收费对账。若收费状态无法确认，准备好的 JSON 会保留在当前运行的 Key-value store，运行停止；请核对平台收费记录后再启动新运行。

Checkpoint recovery and automatic billing reconciliation are not supported. If a charge cannot be confirmed, the prepared JSON remains in the run's Key-value store and the run stops. Check the platform charge records before starting another run.

❓ 常见问题 / Frequently asked questions

可以转写整部红果短剧吗？ / Can I transcribe a complete Hongguo series?

可以。提交 seriesIds，每次最多 10 部剧，展开后不超过 1,000 个唯一单集。每个成功单集单独返回 JSON 并计费。

Yes. Submit seriesIds: up to 10 series and 1,000 unique expanded episodes per run. Each successful episode returns its own JSON result and is priced separately.

用剧集 ID 还是单集 ID 更合适？ / Should I use series IDs or episode IDs?

推荐剧集 ID，因为会携带剧名和简介辅助识别。若只需指定单集，可使用 videoIds；这种模式不会补查所属剧集，部分剧集关联字段可能为 null。

Series IDs are recommended because titles and introductions provide recognition context. Use videoIds for specific episodes; this mode does not look up the parent series, so some association fields may be null.

会识别角色姓名或跨集匹配人物吗？ / Are characters identified across episodes?

不会。speakerId 是集内说话人编号，不是角色身份，也不保证跨集一致。

No. speakerId labels voices within an episode; it is not a character identity and is not guaranteed to match across episodes.

缓存命中后还会收费吗？ / Are cache hits charged?

会。成功交付的缓存结果与新生成结果使用相同事件价格。新运行也会重新计入 Free 账户的月度成功结果配额。

Yes. Successfully delivered cached results have the same event price as newly generated results. A new run also counts toward the Free plan's monthly successful-result cap.

某一集失败会怎样？ / What happens if an episode fails?

Actor 返回该集的状态和错误码，并在预算及配额允许的情况下继续处理其他单集。失败、无语音和未完成项不收成功结果事件费。

The Actor reports that episode's status and error code and continues with other episodes while budget and quota allow. Failed, speech-free, and unfinished items do not incur successful-result event charges.

🤝 支持与使用范围 / Support and permitted use

由 spider_studio 维护，与红果官方无关联。请处理你有权访问和使用的内容。如需帮助，请在 Issues 页提供运行链接，不要公开密钥或私人输入；程序化调用方式见 API 页。

Maintained by spider_studio, independently of Hongguo. Process content you are authorized to access and use. For support, include the run URL in the Issues tab without credentials or private inputs. See the API tab for programmatic integration.

文档更新：2026-11-06。 / Documentation updated: November 6, 2026.