Back to Details

auto-editor-transcribe

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Transcribe & edit by speech

语音转录与内容编辑

Use auto-editor's whisper.cpp (Whisper or NVIDIA Parakeet models) or Apple Speech backend to transcribe audio. Cut a timeline based on subtitle content with 
--edit subtitle
/
word
.
使用auto-editor的whisper.cpp（基于Whisper或NVIDIA Parakeet模型）或Apple Speech后端进行音频转录。通过
--edit subtitle
/
word
参数根据字幕内容剪辑时间线。

Transcribe — 
auto-editor whisper

转录功能 — 
auto-editor whisper

auto-editor whisper <file|:mic> <model> [options]
Set 
<model>
to a 
ggml
model path. Three backends, chosen by the model:
  • Whisper: 
    ggml-small.en.bin
    , 
    ggml-medium.en.bin
    , 
    ggml-large-v3.bin
    , … (https://huggingface.co/ggerganov/whisper.cpp). 99+ languages, 
    --translate
    , 
    --prompt
    .
  • Parakeet: any model with "parakeet" in the filename, e.g. 
    ggml-parakeet-tdt-0.6b-v3-q8_0.bin
    (https://huggingface.co/ggml-org/parakeet-GGUF). Faster than Whisper at comparable English accuracy; language is auto-detected. 
    --translate
    , 
    --prompt
    , and 
    --language
    are rejected.
  • Apple: the magic model name 
    apple
    uses Apple's built-in transcriber; requires macOS 26 or later.
Only the first audio stream of a file is used. Audio is resampled to 16 kHz, and text prints to stdout by default.
bash
auto-editor whisper example.mp4 ggml-medium.en.bin                      # plain text → stdout
auto-editor whisper example.mp4 ggml-medium.en.bin --format srt -o out.srt
auto-editor whisper example.mp4 ggml-parakeet-tdt-0.6b-v3-q8_0.bin      # parakeet backend
auto-editor whisper example.mp4 apple --language en_US                  # macOS 26+
Options: 
--format text|srt|json
, 
-o/--output FILE
, 
-l/--language en
(default auto), 
-tr/--translate
(→ English), 
-sw/--split-words
(one word per cue), 
--queue SECS
(default 30), 
--prompt TEXT
, 
--threads N
(default 4), and 
-t/--threshold THRES
(default 0.04).
--split-words
works with on all backends. Parakeet is the most accurate.
With the 
apple
model, set 
--language
to a supported language or locale. Apple speech cannot auto-detect language, so 
auto
falls back to 
en_US
with a warning. Do not pass 
--translate
or 
--prompt
; neither is supported. The first use of a language may download Apple's speech model and therefore needs network access.
auto-editor whisper <file|:mic> <model> [options]
将
<model>
设置为
ggml
模型路径。根据模型选择三种后端：
仅使用文件中的第一条音频流。音频会重采样至16kHz，默认将文本输出到标准输出。
bash
auto-editor whisper example.mp4 ggml-medium.en.bin                      # 纯文本输出至标准输出
auto-editor whisper example.mp4 ggml-medium.en.bin --format srt -o out.srt
auto-editor whisper example.mp4 ggml-parakeet-tdt-0.6b-v3-q8_0.bin      # 使用Parakeet后端
auto-editor whisper example.mp4 apple --language en_US                  # 适用于macOS 26+
可选参数：
--format text|srt|json
、
-o/--output FILE
、
-l/--language en
（默认自动检测）、
-tr/--translate
（转录并翻译为英文）、
-sw/--split-words
（每个单词单独生成字幕片段）、
--queue SECS
（默认30）、
--prompt TEXT
、
--threads N
（默认4）， 以及
-t/--threshold THRES
（默认0.04）。
--split-words
参数适用于所有后端，其中Parakeet的识别精度最高。
使用
apple
模型时，需将
--language
设置为支持的语言或区域设置。Apple Speech无法自动检测语言，因此设置为
auto
时会默认回退到
en_US
并发出警告。请勿传递
--translate
或
--prompt
参数，二者均不支持。首次使用某一语言时可能需要下载Apple的语音模型，因此需要网络连接。

Transcribe a live microphone

实时麦克风转录

Pass 
:mic
instead of a file. Stop capture gracefully with Ctrl-C.
bash
auto-editor whisper :mic ggml-medium.en.bin
auto-editor whisper :mic apple --language en_US    # macOS 26+
auto-editor whisper :mic ggml-medium.en.bin -o transcript.srt
This streams the microphone directly to transcription and does not save a media recording. To retain and edit the captured audio, use 
auto-editor :mic
; editor and timeline exports save a sibling lossless-FLAC 
_RECORDING.mka
by default.
Live capture supports macOS, Windows, and Linux:
  • macOS uses AVFoundation and prefers a USB microphone, then the system default.
  • Windows uses DirectShow and prefers a USB microphone, then the first audio capture device.
  • Linux uses the default ALSA input device.
When 
-o/--output
ends in 
.srt
, 
.json
, 
.txt
, or 
.text
, the output format is inferred unless 
--format
is set explicitly.
将文件路径替换为
:mic
即可。通过Ctrl-C可优雅停止捕获。
bash
auto-editor whisper :mic ggml-medium.en.bin
auto-editor whisper :mic apple --language en_US    # 适用于macOS 26+
auto-editor whisper :mic ggml-medium.en.bin -o transcript.srt
此功能会将麦克风音频直接流式传输到转录模块，不会保存媒体录制文件。如需保留并编辑捕获的音频，请使用
auto-editor :mic
；编辑器和时间线导出会默认保存一个无损FLAC格式的同级文件
_RECORDING.mka
。
实时捕获支持macOS、Windows和Linux系统：
  • macOS使用AVFoundation优先选择USB麦克风，其次是系统默认麦克风。
  • Windows使用DirectShow优先选择USB麦克风，其次是第一个音频捕获设备。
  • Linux使用默认的ALSA输入设备。
当
-o/--output
的文件名以
.srt
、
.json
、
.txt
或
.text
结尾时，会自动推断输出格式，除非显式设置
--format
参数。

Edit by spoken content

根据语音内容剪辑视频

--edit subtitle
/
word
marks the time a matching subtitle line occupies as active (kept by default). Use an existing subtitle stream, or transcribe to an 
.srt
first and feed it in.
MethodActive when…Args (defaults)
subtitle
/ 
regex
pattern
(regex) matches a line
pattern=""
(empty matches every line), 
stream=0
, 
ignore-case=#f
word
value
appears as a whole word
value
required, 
stream=0
, 
ignore-case=#t
pattern
is optional for 
subtitle
/
regex
: with none, the empty regex matches every subtitle line, so 
--edit subtitle
keeps all sections that have a subtitle (i.e. cut everything with no speech). 
word
, by contrast, requires a 
value
.
bash
undefined
--edit subtitle
/
word
会将匹配字幕片段的时间段标记为保留（默认保留）。可使用已有的字幕流，或先转录生成
.srt
文件再导入使用。
方法保留条件参数（默认值）
subtitle
/ 
regex
pattern
（正则表达式）匹配字幕行
pattern=""
（空值匹配所有行）、
stream=0
、
ignore-case=#f
word
value
作为完整单词出现		必填
value
、
stream=0
、
ignore-case=#t
subtitle
/
regex
的
pattern
参数是可选的：若不设置，空正则表达式会匹配所有字幕行，因此
--edit subtitle
会保留所有有字幕的片段（即删除无语音的静音间隙）。相比之下，
word
方法必须指定
value
参数。
bash
undefined

Keep only sections that have any subtitle (cut the silent gaps between speech)

仅保留有字幕的片段（删除语音之间的静音间隙）

auto-editor talk.mkv --edit subtitle
auto-editor talk.mkv --edit subtitle

Keep only sections where "introduction" is spoken

仅保留提到"introduction"的片段

auto-editor lecture.mp4 --edit word:introduction
auto-editor lecture.mp4 --edit word:introduction

Regex match (case-insensitive); keep matched lines

正则匹配（忽略大小写）；保留匹配的字幕片段

auto-editor talk.mkv --edit "subtitle:pattern=(yes|no),ignore-case=#t"
auto-editor talk.mkv --edit "subtitle:pattern=(yes|no),ignore-case=#t"

Cut filler words instead of keeping them: invert with 
not

删除填充词而非保留：使用
not
反转规则

auto-editor video.mp4 --edit "(not word:um)"
undefined
auto-editor video.mp4 --edit "(not word:um)"
undefined

Transcribe → cut workflow

转录→剪辑工作流

When the media has no subtitle stream, generate a sidecar named after the input — for 
talk.mp4
, auto-editor auto-loads 
talk.srt
(then 
talk.ass
) when the requested subtitle stream isn't embedded:
bash
auto-editor whisper talk.mp4 ggml-medium.en.bin --format srt -o talk.srt
auto-editor talk.mp4 --edit word:question        # picks up talk.srt automatically
Inspect subtitle streams with 
auto-editor subdump FILE
(text subtitles only; bitmap subtitles won't dump).
For loudness/motion-based cutting and pace, see the auto-editor skill.
当媒体文件没有内嵌字幕流时，生成一个与输入文件同名的外置字幕文件——例如对于
talk.mp4
，当请求的字幕流不存在时，auto-editor会自动加载
talk.srt
（若不存在则尝试
talk.ass
）：
bash
auto-editor whisper talk.mp4 ggml-medium.en.bin --format srt -o talk.srt
auto-editor talk.mp4 --edit word:question        # 会自动读取talk.srt
使用
auto-editor subdump FILE
可查看字幕流（仅支持文本字幕；位图字幕无法导出查看）。
如需基于音量/运动的剪辑和节奏调整，请查看auto-editor的核心功能文档。