AWS Textract NodeJS: Detect document from local content as opposed to an S3 url

Question

I have this

    const textract = new AWS.Textract({ region });
    const params = {
        DocumentLocation: {
            S3Object: {
                Bucket: bucketName,
                Name: s3FilePath
            }
        }
    };

    const data = await textract.detectDocumentText(params).promise();

It works fine. However, in another scenario, I do not have this data in S3 url. Instead, I tried

        const params = {
            Document: {
                Bytes: bytes
            }
        };

        const data = await textract.detectDocumentText(params).promise();

I got: UnsupportedDocumentException. It seems mostly I am making wrong API call. Any idea how I can fix error [or] suggestion to use a different API?