Audio & Video to Text

Pricing

$5.00/month + usage

Try for free

Go to Apify Store

Audio & Video to Text

Try for free

Developed by

Donjuan

Maintained by Community

Transcribes video and audio files into plain text and subtitle formats (TXT, SRT, VTT, TSV, JSON) using OpenAI's Whisper model. Supports preloaded tiny, base, and small models.

0.0 (0)

Pricing

$5.00/month + usage

Last modified

3 months ago

Social media

Developer tools

Open source

.actor/README.md

1---
2
3## 🎬 Video and Audio to Text Transcription
4
5### 🧠 Overview
6
7This script is designed for the **Apify** platform and uses **OpenAI Whisper** to transcribe audio or video (e.g., from YouTube or MP4 files) into text and other formats (SRT, VTT, etc.).
8
9---
10
11## 📥 Input
12
13### Parameters
14
15- **model**: *(string)* — Whisper model to use. Available options:
16    - `tiny` ✅ *(pre-installed)*
17    - `base` ✅ *(pre-installed)*
18    - `small` ✅ *(pre-installed)*
19    - `medium` *(requires download)*
20    - `large` *(requires download)*
21    - `turbo` *(requires download)*
22
23> ✅ **Note**: Models `tiny`, `base`, and `small` are already downloaded in the Docker image for faster and offline-ready processing.
24- **source_url**: *(string)* — Direct URL to the video/audio file (e.g., an MP4 file hosted online).  
25  ⚠️ *YouTube links are not supported directly. You must download the video first.*
26
27### Example Input
28
29```json
30{
31  "model": "tiny",
32  "source_url": "https://raw.githubusercontent.com/donjuanMime/audio_to_text/main/video.mp4"
33}
34```
35
36---
37
38## 📤 Output
39
40The output is a JSON array with one object, which includes multiple transcription formats:
41
42* `json`: Full Whisper output with segments, tokens, and metadata.
43* `srt`: SubRip subtitle format.
44* `tsv`: Tab-separated values (start, end, text).
45* `txt`: Plain text transcription.
46* `vtt`: WebVTT subtitle format.
47
48### Example Output (excerpt)
49
50```json
51[
52  {
53    "json": "{ ... Whisper segment data ... }",
54    "srt": "1\n00:00:00,000 --> 00:00:01,120\nWhat's your favorite drink?\n...",
55    "tsv": "start\tend\ttext\n0\t1120\tWhat's your favorite drink?\n...",
56    "txt": "What's your favorite drink?\nMy favorite drink is apple juice...\n",
57    "vtt": "WEBVTT\n\n00:00.000 --> 00:01.120\nWhat's your favorite drink?\n..."
58  }
59]
60```
61
62---
63
64## 🛠️ How to Use
65
661. Go to your **Apify** dashboard and create a new actor or task.
672. Paste this script into the actor’s source.
683. Provide the input in the required JSON format (see above).
694. Run the actor. It will download the media file, process it using Whisper, and return transcription in multiple formats.
70
71---
72
73## ⚠️ Disclaimer
74
75This script is provided **"as is"**, without warranties of any kind. Use it at your own risk.
76Ensure compliance with:
77
78* YouTube’s Terms of Service (if downloading/transcribing from YouTube).
79* Local and international copyright laws.
80
81---
82
83Let me know if you’d like the actual Apify actor code or instructions on downloading YouTube videos as `.mp4` files to use with this.

.actor/actor.json

{
  "actorSpecification": 1,
  "name": "video-to-text",
  "title": "Video to Text",
  "description": "Transcribes video or audio files into text and subtitle formats using Whisper (tiny, base, small models pre-installed).",
  "version": "0.1.0",
  "meta": {
    "templateId": "ts-start-bun"
  },
  "readme": "./README.md",
  "input": "./input_schema.json",
  "dockerfile": "./Dockerfile",
  "storages": {
    "dataset": {
      "actorSpecification": 1,
      "views": {
        "overview": {
          "title": "Overview",
          "transformation": {
            "fields": [
              "json",
              "srt",
              "tsv",
              "txt",
              "vtt"
            ]
          },
          "display": {
            "component": "table",
            "properties": {
              "json": {
                "label": "JSON",
                "format": "text"
              },
              "srt": {
                "label": "SRT",
                "format": "text"
              },
              "tsv": {
                "label": "TSV",
                "format": "text"
              },
              "txt": {
                "label": "Plain Text",
                "format": "text"
              },
              "vtt": {
                "label": "VTT",
                "format": "text"
              }
            }
          }
        }
      }
    }
  }
}

.actor/input_schema.json

{
  "title": "Video Transcription",
  "type": "object",
  "schemaVersion": 1,
  "properties": {
    "source_url": {
      "title": "Source URL",
      "type": "string",
      "description": "Direct URL to the video or audio file (e.g., .mp4, .wav). YouTube links are not supported.",
      "editor": "textfield",
      "prefill": "https://raw.githubusercontent.com/donjuanMime/audio_to_text/main/video.mp4"
    },
    "model": {
      "title": "Whisper Model",
      "type": "string",
      "description": "Whisper model to use: 'tiny', 'base', or 'small' (pre-installed). Other models require download.",
      "editor": "textfield",
      "prefill": "tiny"
    }
  },
  "required": ["source_url", "model"]
}

store/dataset.go

package store

import (
	"bytes"
	"encoding/json"
	"fmt"
	"github.com/davesavic/clink"
	"io"
)

func GetKVDatasetEndpoint(id string) string {
	return fmt.Sprintf(KVDatasetURL, id)
}

func KVDatasetNew(id string) KVDataset {
	client := clink.NewClient()
	client.Headers["Content-Type"] = "application/json"
	return KVDataset{ID: id, Client: client}
}

func KVDatasetDefault() KVDataset {
	return KVDatasetNew(KVDatabaseDefaultID)
}

func (kv KVDataset) Put(payload any) error {
	url := GetKVDatasetEndpoint(kv.ID)
	kv.Client.Headers["Authorization"] = fmt.Sprintf("Bearer %s", Token)
	_, err := kv.Client.Post(url, KVDatasetRequestFrom(payload))
	if err != nil {
		xlog.Error("failed to set dataset")
	}
	return err
}

func KVDatasetRequestFrom(v any) io.Reader {
	b, _ := json.Marshal(v)
	return bytes.NewReader(b)
}

store/file.go

package store

import (
	"bytes"
	"encoding/json"
	"fmt"
	"io"
	"log/slog"
	"net/http"
	"os"
	"strings"

	"github.com/davesavic/clink"
)

var (
	KVStoreDefaultID    string
	KVDatabaseDefaultID string
	KVStoreURL          = "https://api.apify.com/v2/key-value-stores/%s/records/%s?token=%s"
	KVDatasetURL        = "https://api.apify.com/v2/datasets/%s/items"
	Token               string
	xlog                = slog.New(slog.NewTextHandler(os.Stdout, nil))

	Dataset KVDataset
	Store   KVStore
)

type (
	KVStoreValue map[string]any
	KVStore      struct {
		ID     string
		Client *clink.Client
	}
	KVDataset struct {
		ID     string
		Client *clink.Client
	}
)

func init() {
	Token = os.Getenv("APIFY_TOKEN")
	KVStoreDefaultID = os.Getenv("APIFY_DEFAULT_KEY_VALUE_STORE_ID")
	KVDatabaseDefaultID = os.Getenv("APIFY_DEFAULT_DATASET_ID")
	if strings.EqualFold(Token, "") || strings.EqualFold(KVStoreDefaultID, "") {
		xlog.Error("token or default kv store missing")
	}

	Dataset = KVDatasetDefault()
	Store = KVStoreDefault()
}

func GetKVStoreEndpoint(id string, key string) string {
	return fmt.Sprintf(KVStoreURL, id, key, Token)
}

func KVStoreNew(id string) KVStore {
	client := clink.NewClient()
	client.Headers["Content-Type"] = "application/json"
	return KVStore{ID: id, Client: client}
}

func KVStoreDefault() KVStore {
	return KVStoreNew(KVStoreDefaultID)
}

func (kv KVStore) Get(key string) (KVStoreValue, error) {
	url := GetKVStoreEndpoint(kv.ID, key)
	resp, err := kv.Client.Get(url)
	if err != nil {
		xlog.Error("failed to get value", "key", key, "error", err)
		return nil, err
	}
	return KVStoreValueFromResponse(resp), nil
}

func (kv KVStore) Put(key string, payload any) error {
	url := GetKVStoreEndpoint(kv.ID, key)
	_, err := kv.Client.Put(url, KVStoreRequestFrom(payload))
	if err != nil {
		xlog.Error("failed to set value", "key", key, "error", err)
	}
	return err
}

func (kv KVStore) Delete(key string) error {
	url := GetKVStoreEndpoint(kv.ID, key)
	_, err := kv.Client.Delete(url)
	if err != nil {
		xlog.Error("failed to delete key", "key", key, "error", err)
	}
	return err
}

func KVStoreRequestFrom(v any) io.Reader {
	b, _ := json.Marshal(v)
	return bytes.NewReader(b)
}

func KVStoreValueFromResponse(resp *http.Response) KVStoreValue {
	var value KVStoreValue
	if err := clink.ResponseToJson(resp, &value); err != nil {
		xlog.Error("failed to unmarshal response", "error", err)
	}
	return value
}

Dockerfile

FROM theartist497/whisper-go-app:latest

# Upgrade pip and install whisper-ctranslate2
RUN pip install --upgrade pip \
    && pip install -U whisper-ctranslate2

# Set working directory inside container
WORKDIR /app

# Copy your app files
COPY . .

# Build Go binary
RUN go build -o main .

# Run the app
CMD ["./main"]

go.mod

module audio_to_text

go 1.23.0

require (
	github.com/davesavic/clink v1.0.2 // indirect
	golang.org/x/time v0.5.0 // indirect
)

go.sum

github.com/davesavic/clink v1.0.2 h1:zpbyoWtx9ztI6nO4wHS72Du6aS3xvtFvdhwn8KOeyh8=
github.com/davesavic/clink v1.0.2/go.mod h1:t/riPX5tWlWKYcP7lvUpfsxFqC6WNnOOO+xvgei+uhg=
golang.org/x/time v0.5.0 h1:o7cqy6amK/52YcAKIPlM3a+Fpj35zvRj2TP+e1xFSfk=
golang.org/x/time v0.5.0/go.mod h1:3BpzKBy/shNhVucY/MWOyx10tF3SFh9QdLuxbVysPQM=

main.go

package main

import (
	"audio_to_text/store"
	"bytes"
	"fmt"
	"io"
	"log"
	"log/slog"
	"net/http"
	"os"
	"os/exec"
	"strings"
	"sync"
)

var (
	xlog = slog.New(slog.NewTextHandler(os.Stdout, nil))
)

func main() {
	input, err := store.Store.Get("INPUT")
	if err != nil {
		xlog.Error("failed to get input from kv store", "error", err)
		return
	}

	source := input["source_url"].(string)
	if strings.EqualFold(source, "") {
		xlog.Error("not found 'source_url'", source)
	}

	model := input["model"].(string)
	if strings.EqualFold(source, "") {
		xlog.Error("not found 'source_url'", source)
	}

	transcription(source, model)
}

func Test() {
	url := "https://raw.githubusercontent.com/donjuanMime/audio_to_text/main/video.mp4"
	model := "tiny"
	transcription(url, model)
}

func transcription(sourceUrl, model string) {
	downloadFile(sourceUrl)
	filename := "video.mp4"

	command := fmt.Sprintf(`whisper-ctranslate2 %s --batched true --batch_size 4 --max_words_per_line 8 --word_timestamps true --vad_filter true --model %s`, filename, model)

	runCmdWithProgress(command)
}

func runCmdWithProgress(command string) bool {
	cmd := exec.Command("bash", "-c", command)

	var stdoutBuf, stderrBuf bytes.Buffer
	stdoutIn, _ := cmd.StdoutPipe()
	stderrIn, _ := cmd.StderrPipe()

	var errStdout, errStderr error
	stdout := io.MultiWriter(os.Stdout, &stdoutBuf)
	stderr := io.MultiWriter(os.Stderr, &stderrBuf)
	if err := cmd.Start(); err != nil {
		xlog.Error("cmd.Start() failed with ", err)
		return false
	}

	var wg sync.WaitGroup
	wg.Add(1)

	go func() {
		_, errStdout = io.Copy(stdout, stdoutIn)

		log.Println(stdoutIn)
		wg.Done()
	}()

	_, errStderr = io.Copy(stderr, stderrIn)
	wg.Wait()

	if err := cmd.Wait(); err != nil {
		xlog.Error("cmd.Run() failed with ", err)
	}
	if errStdout != nil || errStderr != nil {
		xlog.Error("failed to capture stdout or stderr", errStderr, errStdout)
		return false
	}

	sendResult()

	return true
}

func sendResult() {
	if err := store.Dataset.Put(map[string]interface{}{
		"json": string(mustReadFile("video.json")),
		"srt":  string(mustReadFile("video.srt")),
		"tsv":  string(mustReadFile("video.tsv")),
		"txt":  string(mustReadFile("video.txt")),
		"vtt":  string(mustReadFile("video.vtt")),
	}); err != nil {
		log.Fatal(err)
	}
}

func mustReadFile(filename string) []byte {
	file, err := os.ReadFile(filename)
	if err != nil {
		return nil
	}
	return file
}

func downloadFile(fileURL string) {
	outputPath := "video.mp4"

	// Create the file
	out, err := os.Create(outputPath)
	if err != nil {
		log.Fatalf("Failed to create file: %v", err)
	}
	defer out.Close()

	// Get the data
	resp, err := http.Get(fileURL)
	if err != nil {
		log.Fatalf("Failed to download file: %v", err)
	}
	defer resp.Body.Close()

	// Check server response
	if resp.StatusCode != http.StatusOK {
		log.Fatalf("Bad status: %s", resp.Status)
	}

	// Write the body to file
	_, err = io.Copy(out, resp.Body)
	if err != nil {
		log.Fatalf("Failed to save file: %v", err)
	}

	log.Println("File downloaded successfully:", outputPath)
}

Audio and Video Transcript (OpenAI Whisper)

vittuhy/audio-and-video-transcript

This Actor transcribes audio or video files from publicly accessible URLs using OpenAI's Whisper API. To use this Actor, you'll need to provide your own OpenAI API key. It supports multiple languages and highly customizable parameters, enabling precise control over the transcription process.

Vít Tuhý

1.8

Audio And Video Transcriber (OpenAI GPT-4o-transcribe)

stanvanrooy6/audio-video-transcriber

Downloads videos from public URLs, extracts audio, and transcribes them using OpenAI

Stan Van Rooy

5.0

Speech to Text Converter (Transcript / Captcha)

saswave/speech-to-text-converter

Transform audio records to text. Get transcription from sales or customer success teams audio files. Get Captcha text from captcha audio challenge. Speech to text converter helps you analyse, build KPI with audio records and bypass captcha.

SASWAVE

Video to Text Transcription

aizen0/video-to-text-transcription

Convert video speech to text in bulk. Supports Only Twitter/Instagram, auto-detects languages, handles large files automatically. Uses OpenAI Whisper for high accuracy.

Pratham Yadav

Text-to-Speech Generator (OpenAI voice generator)

stanvanrooy6/text-to-speech-generator-openai-voice-generator

Convert text to speech effortlessly with our OpenAI voice generator. Choose from 6 English-optimized voices, customize settings, and get high-quality audio files fast. Simple to use, integrates with your OpenAI API key.

Stan Van Rooy

1.9

Universal Speech to Text Transcriber

tictechid/vanzi-universal-transcriber

Transcribe audio from videos stored on Google Drive, Dropbox, GitHub raw, OneDrive, Box, iCloud, AWS S3, GCS, Azure Blob, and Backblaze B2. Convert share links to direct downloads for fast, accurate transcripts with timestamps and easy API integration.

TicTech

5.0

📹Video to Text

nextapi/video-to-text

Easily convert videos to text from over 1000 platforms, including YouTube, TikTok, Twitter/X, Instagram... Supports 12+ languages with translation options.

NextAPI

3.4

📟Live Stream Transcript - Real-Time Video/Audio Transcription

agentx/live-stream-transcript

Experience the future of live streaming with our AI-powered Transcript API. Capture, transcribe, and analyze streams from 1000+ platforms in real-time!