AWS Lambda JSON 格式化

Question

I am trying to change how the output of Amazon Transcribe looks and I am doing this through a Lambda. The Transcribe output produces a JSON file with the appropriate information, which is then stored in an S3 bucket. This then triggers another lambda which will format it accordingly.

Currently, the JSON file can be found at the following link (it is too large to post on here: https://paste.pythondiscord.com/ebenugoqok.py

And I am trying to get the Transcript so it looks like this:

ch_1 :  just
ch_0 :  Hey,  Steve,  this is Mike calling ARM.  How you doing this afternoon? 
ch_1 :  I'm alright.  How are you doing? 
ch_0 :  Um,  doing well.  So,  uh,  ARM just have me reach out here today to update things on our end and make sure that we're keeping you guys up to date with the latest pricing specs and technology for 2020.  Um,  just looking to see if you're more focused on,  uh,  networking or data center initiatives right now. 
ch_1 :  Ah,  either be crazy.  But the
ch_0 :  Yeah,  I was just doing a great I gotcha.  I gotcha.  And,  um are you guys currently working with a ARM partner? 
ch_1 :  Yeah Oh yeah, 
ch_0 :  You're right. 
ch_1 :  That was one of the people that thing, 
ch_0 :  Okay.  Alright.  And,  uh,  just like an update the notes here on my end.  Um,  do you guys have any projects that are slated for,  like,  the next 12 months?  Right? 
ch_1 :  but
ch_0 :  Crap.  Okay.  Alright.  I will,  uh,  go ahead and update the nose when I went to reflect that.  Ah,  I do appreciate your time. 
ch_1 :  Alright,  thank you very much.  You have to go. 
ch_0 :  Bye.  Did you pick

I have the following code in my lambda:

import json
import boto3

def lambda_handler(event, context):
    if event:
        s3 = boto3.client("s3")
        s3_object = event["Records"][0]["s3"]
        bucket_name = s3_object["bucket"]["name"]
        file_name = s3_object["object"]["key"]
        file_obj = s3.get_object(Bucket=bucket_name, Key=file_name)
        transcript_result = json.loads(file_obj["Body"].read())

        channels = transcript_result["results"]["channel_labels"]
        items = transcript_result["results"]["items"]


        speaker_text = []
        flag = False

        temp = None


        with open("/tmp/transcribe.txt", "a") as x:
            for word in items:
                for seg in channels["channels"]:
                    for seg_item in seg["items"]:
                        # print(word["type"])
                        if "start_time" in seg_item and word["type"] != "punctuation":
                            if word["end_time"] == seg_item["end_time"] and word["start_time"] == seg_item["start_time"]:
                                # if word["alternatives"][0]["content"]:
                                if temp != seg["channel_label"]:
                                    x.write("\n")
                                    x.write("{} : ".format(seg["channel_label"]))
                                    speaker_text.append(word["alternatives"][0]["content"])
                                    flag = True
                                    temp = seg["channel_label"]
                                else:
                                    speaker_text.append(word["alternatives"][0]["content"])
                                    flag = True

            if word["type"] == "punctuation":
                x.write(word["alternatives"][0]["content"])
            x.write(" {}".format(' '.join(speaker_text)))

    s3.put_object(Bucket="aws-channel-separation", Key=file_name, Body=json.dumps(speaker_text))

However, the output of this is:

["just", "Hey", "Steve", "this", "is", "Mike", "calling", "ARM", "How", "you", "doing", "this", "afternoon", "I'm", "alright", "How", "are", "you", "doing", "Um", "doing", "well", "So", "uh", "ARM", "just", "have", "me", "reach", "out", "here", "today", "to", "update", "things", "on", "our", "end", "and", "make", "sure", "that", "we're", "keeping", "you", "guys", "up", "to", "date", "with", "the", "latest", "pricing", "specs", "and", "technology", "for", "2020", "Um", "just", "looking", "to", "see", "if", "you're", "more", "focused", "on", "uh", "networking", "or", "data", "center", "initiatives", "right", "now", "Ah", "either", "be", "crazy", "But", "the", "Yeah", "I", "was", "just", "doing", "a", "great", "I", "gotcha", "I", "gotcha", "And", "um", "are", "you", "guys", "currently", "working", "with", "a", "ARM", "partner", "Yeah", "Oh", "yeah", "You're", "right", "That", "was", "one", "of", "the", "people", "that", "thing", "Okay", "Alright", "And", "uh", "just", "like", "an", "update", "the", "notes", "here", "on", "my", "end", "Um", "do", "you", "guys", "have", "any", "projects", "that", "are", "slated", "for", "like", "the", "next", "12", "months", "Right", "but", "Crap", "Okay", "Alright", "I", "will", "uh", "go", "ahead", "and", "update", "the", "nose", "when", "I", "went", "to", "reflect", "that", "Ah", "I", "do", "appreciate", "your", "time", "Alright", "thank", "you", "very", "much", "You", "have", "to", "go", "Bye", "Did", "you", "pick"]

Does anyone know how to get the formatting correct? Any help would be appreciated.

Answer 1

On this line, you have speaker_text formatted as JSON and stored in S3:

s3.put_object(Bucket="aws-channel-separation", Key=file_name, Body=json.dumps(speaker_text))

But it seems like you'd actually want the text that was written to /tmp/transcribe.txt . You can do that with:

s3.upload_file(Filename="/tmp/transcribe.txt", Bucket="aws-channel-separation", Key=file_name)