Unable to select Audio and Video

Google AI Studio
1

I am unable to select a m3a or mp4 or any other media format file. It says file type not supported.

How do I try out MultiModal capability?

2

Welcome to the forums!

I’ve given AI Studio an mp4 file several times without any problems. How long is the file, and are you sure it’s a real mp4 file?

A screen shot illustrating the problem may help us help you figure out what’s going on.

3

Here is the screenshot

This is simple 4 second audio file with. I have also tried with mp4 and other audio files.
Uploading it to drive and selecting files from the drive is also attempted.

Screenshot 2024-07-16 at 2.37.41 PM
Screenshot 2024-07-16 at 2.37.41 PM2810×2014 349 KB

4

Welcome to the Forum. I will hazard a guess that you made the recording using an iPhone or iPad. The google code doesn’t understand that the .m4a extension is a perfectly fine audio file. I had the same problem with my testing and wrote a helper utility that is used like this

part_util.makeBlobPart('REST/Econ_cycles.m4a', mimetype="audio/mp4")

In other words, by specifying the mime type the API will accept and process the file. The code for the utility looks like this

def makeBlobPart(filename:str, mimetype: str = None) -> Part:
    # REST/image.jpg
    return Part(inline_data=makeBlob(filename, mimetype))

def makeBlob(filename:str, mimetype: str = None) -> Blob:
    blobby = make_blob(filename, mimetype)
    return Blob(**blobby)

def make_blob(filename:str, mimetype: str = None) -> dict:
    # Test image has "totalTokens": 262
    blob_part = dict()
    if mimetype == None:
        guess = mimetypes.guess_type(filename)
        if guess != None:
            logger.info("make_blob: " + filename + " guessed mimetype " 
                        + guess[0])
            blob_part.update({"mime_type": guess[0]})
        else:
            logger.warn("make_blob: " + filename + " cannot guess mimetype")
            raise TypeError # can not reasonably proceed
    else:
        logger.info("make_blob: " + filename + " given mimetype " + mimetype)
        blob_part.update({"mime_type": mimetype})
    with open(filename, 'rb') as image_file:
        base64data = base64.standard_b64encode(image_file.read())
        blob_part.update({"data": base64data.decode('utf-8')})
    return blob_part

Hope that helps.

1 Like