Recherche avancée

Sur d’autres sites (12049)

• How do I deploy Whisper.cpp stream.wasm on an EC2 instance ? [closed]

  3 août, par tosUser

  How do I deploy the whisper.cpp stream.wasm demo on a EC2 instance ?

  


  The demo is available from the following git :
https://github.com/ggml-org/whisper.cpp/tree/master/examples/stream.wasm

  


  I am fairly far along, but the program hangs on Preparing ...
I do not know what is causing the issue.

  


  Here are the steps I have taken. (I am trying to make it brief, and provide more detail as necessary.)
I cloned the repository and built it locally.
It runs properly.

  


  I created an EC2 instance running Amazon Linux 2023, and uploaded the Whisper.cpp files to it.
I ssh into the server and installed nginx, it displays the default page to the server's public ip address.
I then configured nginx as a reverse proxy server :

  


      server {
        listen 80;
        server_name redactedPublicIpAddress;
        location / {
            proxy_pass http://127.0.0.1:8000;
        }
    }



  


  I enabled and restarted nginx.

  


  I then try to run whisper.cpp stream.wasm by running python3 examples/server.py
It indicates that it is serving home/ec2-user/whisper.cpp/build-em/bin to localhost:8000
This is analogous to the message when it is served locally.
Stream.wasm has a 301 response.
Helpers.js and coi-serviceworker.js have 200 responses.

  


  The index file displays normally at the public ip address/stream.wasm/

  


  However, all is not well. The js seems to hang on Preparing...
Other js functions on the page still function, I can download a model and the page indicates it was downloaded, etc. However, I cannot start recording. The start button is not active. It never properly initializes.

  


  This type of behavior is identical to trying to open the index file on my local computer without it being served.

  


  I think I am missing something simple, but do not know what it is.

  


  Here are some steps that I took afterwards that did not fix the issue :
I installed nodejs and npm so that I could install ffmpeg.wasm : npm install @ffmpeg/ffmpeg @ffmpeg/util

  


  The github page indicates that I need to put the files in the html path. So in trying to troubleshoot I put them in /user/shared/nginx/html which is where the nginx default index file is placed.

  


  I think it may be related to serving the python. The server.py file is not in a state to deploy it with uvicorn.

  


  Maybe there is some sort of cors issue.

  


  I simply do not know.

  


• "Invalid or unexpected token" error when trying to execute ffmpeg build on lambda

  4 janvier 2019, par almarc

  I have a node.js script that uses ffmpeg to convert mp4 downloaded from YT to mp3 and save to Amazon S3. Uploading using the serverless framework. The "ffmpeg" file is included in the main directory (with .yml), downloaded from here :
  https://johnvansickle.com/ffmpeg/

  The code :

  'use strict'
  
  process.env.PATH = process.env.PATH + ':/tmp/'
  
  process.env['FFMPEG_PATH'] = '/tmp/ffmpeg';
  
  const BIN_PATH = process.env['LAMBDA_TASK_ROOT']
  
  process.env['PATH'] = process.env['PATH'] + ':' + BIN_PATH;
  
  
  
  module.exports.download_mp3 = function (event, context, callback)
  
  {
  
    require('child_process').exec('cp /var/task/ffmpeg /tmp/.; chmod 755 
  
    /tmp/ffmpeg;', function (error, stdout, stderr) {
  
    if (error) 
  
    {
  
      console.log('An error occured', error);
  
      callback(null, null)
  
    } 
  
    else 
  
    {
  
      var ffmpeg = require('ffmpeg');
  
      const aws = require('aws-sdk')
  
      const s3 = new aws.S3()
  
      const ytdl = require('ytdl-core');
  
  
  
      function uploadFromStream(s3) {
  
        const stream = require('stream')
  
        var pass = new stream.PassThrough();
  
  
  
        var params = {Bucket: "some-bucket", Key: "some-key", Body: pass};
  
        s3.upload(params, function(err, data) {
  
          console.log(err, data);
  
        });
  
        console.log("Should be finished")
  
        callback(null)
  
      }
  
  
  
      let stream = ytdl("some-video-id", {
  
        quality: 'highestaudio',
  
        filter: 'audioonly'
  
      });
  
  
  
      ffmpeg(stream)
  
        .audioBitrate(128)
  
        .format('mp3')
  
        .on('error', (err) => console.error(err))
  
        .pipe(uploadFromStream(s3), {
  
          end: true
  
      });
  
    }})
  
  }

  When triggered, the function writes an error in logs :
  

  2019-01-04T14:50:54.525Z    21da4d49-1030-11e9-b901-0dc32b691a16     
  
  /var/task/ffmpeg:1
  
  (function (exports, require, module, __filename, __dirname) { ELF
  
  ^
  
  
  
  SyntaxError: Invalid or unexpected token
  
  at createScript (vm.js:80:10)
  
  at Object.runInThisContext (vm.js:139:10)
  
  at Module._compile (module.js:616:28)
  
  at Object.Module._extensions..js (module.js:663:10)
  
  at Module.load (module.js:565:32)
  
  at tryModuleLoad (module.js:505:12)
  
  at Function.Module._load (module.js:497:3)
  
  at Module.require (module.js:596:17)
  
  at require (internal/module.js:11:18)
  
  at /var/task/download.js:17:18

  It’s, most definetely, an error in the "ffmpeg" file I’ve mentioned above (link provided). But I don’t know what’s the exact issue, I followed the first answer here : https://stackoverflow.com/questions/47882810/lambda-not-connecting-to-ffmpeg to include the ffmpeg build.

• FFmpeg transcoding on Lambda results in unusable (static) audio

  17 mai 2020, par jmkmay

  I'd like to move towards serverless for audio transcoding routines in AWS. I've been trying to setup a Lambda function to do just that ; execute a static FFmpeg binary and re-upload the resulting audio file. The static binary I'm using is here.

  



  The Lambda function I'm using in Python looks like this :

  



  import boto3

s3client = boto3.client('s3')
s3resource = boto3.client('s3')

import json
import subprocess 

from io import BytesIO

import os

os.system("cp -ra ./bin/ffmpeg /tmp/")
os.system("chmod -R 775 /tmp")

def lambda_handler(event, context):

    bucketname = event["Records"][0]["s3"]["bucket"]["name"]
    filename = event["Records"][0]["s3"]["object"]["key"]

    audioData = grabFromS3(bucketname, filename)

    with open('/tmp/' + filename, 'wb') as f:
        f.write(audioData.read())

    os.chdir('/tmp/')

    try:
        process = subprocess.check_output(['./ffmpeg -i /tmp/joe_and_bill.wav /tmp/joe_and_bill.aac'], shell=True, stderr=subprocess.STDOUT)
        pushToS3(bucketname, filename)
        return process.decode('utf-8')
    except subprocess.CalledProcessError as e:
        return e.output.decode('utf-8'), os.listdir()


def grabFromS3(bucket, file):

    obj = s3client.get_object(Bucket=bucket, Key=file)
    data = BytesIO(obj['Body'].read())

    return(data)

def pushToS3(bucket, file):

    s3client.upload_file('/tmp/' + file[:-4] + '.aac', bucket, file[:-4] + '.aac')

    return


  



  You can listen to the output of this here. WARNING : Turn your volume down or your ears will bleed.

  



  The original file can be heard here.

  



  Does anyone have any idea what might be causing the encoding errors ? It doesn't seem to be an issue with the file upload, since the md5 on the Lambda fs matches the MD5 of the uploaded file.

  



  I've also tried building the static binary on an Amazon Linux instance in EC2, then zipping and porting it into the Lambda project, but the same issue persists.

  



  I'm stumped ! :(