I don't know what is option's argument

[blusewill]

So I am writing a python script to generate SRT file
and I don't know how it's won't available to generate the file
It's will keep saying TypeError: call() missing 1 required positional argument: 'options' and I checked every single forum thread but there is no similar one that I encountered, hope someone can help me to fix this issue

[phineas-pta]

this is a new feature in development, you should instead use the stable version, in colab, rather than

pip install git+https://github.com/openai/whisper.git

you do

pip install openai-whisper

[g-l-i-t-c-h-o-r-s-e]

this is a new feature in development, you should instead use the stable version, in colab, rather than

pip install git+https://github.com/openai/whisper.git

you do

pip install openai-whisper

What is the options argument though?

[erkkimon]

I'm wondering the same. It would be nice to know how to use the latest version, because I bet it's not really broken.

[erkkimon]

After taking a further look, it really seems broken. Looks like the options will do stuff in the future but not yet, so many functions expect options but internally it isn't passed, so you can't pass them to the expecting functions without modifying the source code.

Here's an ugly fix which works for JSON format. You can delete options for other writing functions as well if you wish. This works if you call get_writer like this this:

writer = get_writer("json", "./output")

And here comes the ugly fix in case someone needs.

--- utils.py.original	2023-09-21 15:29:45.437190286 +0300
+++ utils.py.modified	2023-09-21 15:28:54.433188605 +0300
@@ -74,7 +74,7 @@
     def __init__(self, output_dir: str):
         self.output_dir = output_dir
 
-    def __call__(self, result: dict, audio_path: str, options: dict):
+    def __call__(self, result: dict, audio_path: str):
         audio_basename = os.path.basename(audio_path)
         audio_basename = os.path.splitext(audio_basename)[0]
         output_path = os.path.join(
@@ -82,9 +82,9 @@
         )
 
         with open(output_path, "w", encoding="utf-8") as f:
-            self.write_result(result, file=f, options=options)
+            self.write_result(result, file=f)
 
-    def write_result(self, result: dict, file: TextIO, options: dict):
+    def write_result(self, result: dict, file: TextIO):
         raise NotImplementedError
 
 
@@ -231,7 +231,7 @@
 class WriteJSON(ResultWriter):
     extension: str = "json"
 
-    def write_result(self, result: dict, file: TextIO, options: dict):
+    def write_result(self, result: dict, file: TextIO):
         json.dump(result, file)
 
 
@@ -249,9 +249,9 @@
     if output_format == "all":
         all_writers = [writer(output_dir) for writer in writers.values()]
 
-        def write_all(result: dict, file: TextIO, options: dict):
+        def write_all(result: dict, file: TextIO):
             for writer in all_writers:
-                writer(result, file, options
)
+                writer(result, file, {})
 
         return write_all

PS. I added a modified utils.py as attachments and a working example of working call as attachments.

utils.py.modified.txt
test.py.txt

[bradsec]

Had the same issue trying to integrate into a python program and save different output format types .srt .json etc - even using pip install git+https://github.com/openai/whisper.git --upgrade. Running as a CLI seemed to work fine, however trying to run like in the below code, regardless of file type the options error always displayed TypeError: ResultWriter.__call__() missing 1 required positional argument: 'options'

def whisper_transcription(input_audio, whisper_model):

    model = whisper.load_model(whisper_model)
    result = model.transcribe(input_audio, verbose=True)
    output_directory = "."

    txt_writer = get_writer("txt", output_directory)
    txt_writer(result, input_audio)

    json_writer = get_writer("json",  output_directory)
    json_writer(result, input_audio)

    vtt_writer = get_writer("vtt", output_directory)
    vtt_writer(result, input_audio)

    tsv_writer = get_writer("tsv", output_directory)
    tsv_writer(result, input_audio)

    srt_writer = get_writer("srt", output_directory)
    srt_writer(result, input_audio)

Error seemed to be caused by this section of the Whisper utils.py when no values or default values were being received by some of the options -

class SubtitlesWriter(ResultWriter):
    always_include_hours: bool
    decimal_marker: str

    def iterate_result(self, result: dict, options: dict):

        raw_max_line_width: Optional[int] = options["max_line_width"] # This line
        max_line_count: Optional[int] = options["max_line_count"] # This line
        highlight_words: bool = options["highlight_words"] # This line
        max_line_width = 1000 if raw_max_line_width is None else raw_max_line_width
        preserve_segments = max_line_count is None or raw_max_line_width is None

As a workaround I made the following changes to my application, and passed in some default options values.

def whisper_transcription(input_audio, whisper_model):

    model = whisper.load_model(whisper_model)
    result = model.transcribe(input_audio, verbose=True)
    output_directory = "."

    # Set some initial options values
    options = {
        'max_line_width': None,
        'max_line_count': None,
        'highlight_words': False
    }

   # Include options in call
    txt_writer = get_writer("txt", output_directory)
    txt_writer(result, input_audio, options)

    json_writer = get_writer("json",  output_directory)
    json_writer(result, input_audio, options)

    vtt_writer = get_writer("vtt", output_directory)
    vtt_writer(result, input_audio, options)

    tsv_writer = get_writer("tsv", output_directory)
    tsv_writer(result, input_audio, options)

    srt_writer = get_writer("srt", output_directory)
    srt_writer(result, input_audio, options)

Another option may just be to construct args in python and use the CLI whisper command formatting.

[blusewill]

Never Thought I am going to open this discussion again.
Because it happen again NOT ON GITHUB RELEASE but actually stable release

The Code can be found at Here https://github.com/blusewill/ytvideo-whisper/tree/dev

[glangford]

@blusewill There was a breaking change made to this API. @Mdsajal5 and @bradsec already documented workarounds above so this is just another way of solving the problem.

Depending on your need you can either call

writer( result, file, {} )

or with the complete set of options, for SRT files for example:

writer(result, file, {"max_line_width":55, "max_line_count":2, "highlight_words":False} ) 

The options dictionary is checked in the first few lines of this code in whisper, fyi.

whisper/whisper/utils.py

Line 103 in 0a60fca

def iterate_result(self, result: dict, options: dict):

[BigApe99]

this error went away when I installed a certain version of whisper:
pip install openai-whisper==20230314

[blusewill]

@glangford Thanks for Mentioning. I am currently testing these options and see it will work in my code or not

[mlane3]

these directions are not clear. I am going to make a new support ticket as a bug.