Create Transcription (Whisper) with OpenAI (ChatGPT) API on New Inbound Message from SMSTools API

Pipedream makes it easy to connect APIs for OpenAI (ChatGPT), SMSTools and 2,400+ other apps remarkably fast.

Trigger workflow on

New Inbound Message from the SMSTools API

Next, do this

Create Transcription (Whisper) with the OpenAI (ChatGPT) API

No credit card required

▶

Watch us build a workflow

8 min

Watch now ➜

Trusted by 1,000,000+ developers from startups to Fortune 500 companies

Developers ♥ Pipedream

Getting Started#

This integration creates a workflow with a SMSTools trigger and OpenAI (ChatGPT) action. When you configure and deploy the workflow, it will run on Pipedream's servers 24x7 for free.

Select this integration
Configure the New Inbound Message trigger
1. Connect your SMSTools account
2. Configure timer
Configure the Create Transcription (Whisper) action
1. Connect your OpenAI (ChatGPT) account
2. Select a Audio Upload Type
3. Optional- Select a Language
Deploy the workflow
Send a test event to validate your setup
Turn on the trigger

Details#

This integration uses pre-built, source-available components from Pipedream's GitHub repo. These components are developed by Pipedream and the community, and verified and maintained by Pipedream.

To contribute an update to an existing component or create a new component, create a PR on GitHub. If you're new to Pipedream component development, you can start with quickstarts for trigger span and action development, and then review the component API reference.

Trigger#

New Inbound Message on SMSTools

Description:Emit new event when a new inbound message is received.

Version:0.0.1

Key:smstools-new-inbound-message

View on GitHub

Trigger Code#

import { DEFAULT_POLLING_SOURCE_TIMER_INTERVAL } from "@pipedream/platform";
import smstools from "../../smstools.app.mjs";
import sampleEmit from "./test-event.mjs";

export default {
  key: "smstools-new-inbound-message",
  name: "New Inbound Message",
  description: "Emit new event when a new inbound message is received.",
  version: "0.0.1",
  type: "source",
  dedupe: "unique",
  props: {
    smstools,
    db: "$.service.db",
    timer: {
      type: "$.interface.timer",
      default: {
        intervalSeconds: DEFAULT_POLLING_SOURCE_TIMER_INTERVAL,
      },
    },
  },
  methods: {
    _getLastId() {
      return this.db.get("lastId") || 0;
    },
    _setLastId(lastId) {
      this.db.set("lastId", lastId);
    },
    async emitEvent(maxResults = false) {
      const lastId = this._getLastId();
      const response = this.smstools.paginate({
        fn: this.smstools.getInboxMessages,
        maxResults,
      });

      let responseArray = [];
      for await (const item of response) {
        if (item.ID <= lastId) break;
        responseArray.push(item);
      }

      if (responseArray.length) {
        this._setLastId(responseArray[0].ID);
      }

      for (const item of responseArray.reverse()) {
        this.$emit(item, {
          id: item.ID,
          summary: `New inbound message from ${item.sender}`,
          ts: Date.parse(item.date),
        });
      }
    },
  },
  hooks: {
    async deploy() {
      await this.emitEvent(25);
    },
  },
  async run() {
    await this.emitEvent();
  },
  sampleEmit,
};

Trigger Configuration#

This component may be configured based on the props defined in the component code. Pipedream automatically prompts for input values in the UI and CLI.

Label	Prop	Type	Description
SMSTools	`smstools`	`app`	This component uses the SMSTools app.
N/A	`db`	`$.service.db`	This component uses `$.service.db` to maintain state between executions.
	`timer`	`$.interface.timer`

Trigger Authentication#

SMSTools uses API keys for authentication. When you connect your SMSTools account, Pipedream securely stores the keys so you can easily authenticate to SMSTools APIs in both code and no-code steps.

To retrieve your Client ID and Client Secret,

Navigate to your SMSTools account and sign in
Go to “Developers” > “API & Webhooks”

About SMSTools#

Text Message Marketing Platform Low cost Bulk SMS

Action#

Create Transcription (Whisper) on OpenAI (ChatGPT)

Description:Transcribes audio into the input language. [See the documentation](https://platform.openai.com/docs/api-reference/audio/create).

Version:0.1.14

Key:openai-create-transcription

View on GitHub

OpenAI (ChatGPT) Overview#

OpenAI provides a suite of powerful AI models through its API, enabling developers to integrate advanced natural language processing and generative capabilities into their applications. Here’s an overview of the services offered by OpenAI's API:

Use Python or Node.js code to make fully authenticated API requests with your OpenAI account:

Action Code#

import ffmpegInstaller from "@ffmpeg-installer/ffmpeg";
import { ConfigurationError } from "@pipedream/platform";
import axios from "axios";
import Bottleneck from "bottleneck";
import { exec } from "child_process";
import FormData from "form-data";
import fs from "fs";
import {
  extname,
  join,
} from "path";
import stream from "stream";
import { promisify } from "util";
import openai from "../../openai.app.mjs";
import common from "../common/common.mjs";
import constants from "../../common/constants.mjs";
import lang from "../common/lang.mjs";

const COMMON_AUDIO_FORMATS_TEXT = "Your audio file must be in one of these formats: mp3, mp4, mpeg, mpga, m4a, wav, or webm.";
const CHUNK_SIZE_MB = 20;

const execAsync = promisify(exec);
const pipelineAsync = promisify(stream.pipeline);

export default {
  name: "Create Transcription (Whisper)",
  version: "0.1.14",
  key: "openai-create-transcription",
  description: "Transcribes audio into the input language. [See the documentation](https://platform.openai.com/docs/api-reference/audio/create).",
  type: "action",
  props: {
    openai,
    uploadType: {
      label: "Audio Upload Type",
      description: "Are you uploading an audio file from [your workflow's `/tmp` directory](https://pipedream.com/docs/code/nodejs/working-with-files/#the-tmp-directory), or providing a URL to the file?",
      type: "string",
      options: [
        "File",
        "URL",
      ],
      reloadProps: true,
    },
    language: {
      label: "Language",
      description: "**Optional**. The language of the input audio. Supplying the input language will improve accuracy and latency.",
      type: "string",
      optional: true,
      options: lang.LANGUAGES.map((l) => ({
        label: l.label,
        value: l.value,
      })),
    },
  },
  async additionalProps() {
    const props = {};
    switch (this.uploadType) {
    case "File":
      props.path = {
        type: "string",
        label: "File Path",
        description: `A path to your audio file to transcribe, e.g. \`/tmp/audio.mp3\`. ${COMMON_AUDIO_FORMATS_TEXT} Add the appropriate extension (mp3, mp4, etc.) on your filename — OpenAI uses the extension to determine the file type. [See the Pipedream docs on saving files to \`/tmp\`](https://pipedream.com/docs/code/nodejs/working-with-files/#writing-a-file-to-tmp)`,
      };
      break;
    case "URL":
      props.url = {
        type: "string",
        label: "URL",
        description: `A public URL to the audio file to transcribe. This URL must point directly to the audio file, not a webpage that links to the audio file. ${COMMON_AUDIO_FORMATS_TEXT}`,
      };
      break;
    default:
      throw new ConfigurationError("Invalid upload type specified. Please provide 'File' or 'URL'.");
    }
    // Because we need to display the file or URL above, and not below, these optional props
    // TODO: Will be fixed when we render optional props correctly when used with additionalProps
    props.prompt = {
      label: "Prompt",
      description: "**Optional** text to guide the model's style or continue a previous audio segment. The [prompt](https://platform.openai.com/docs/guides/speech-to-text/prompting) should match the audio language.",
      type: "string",
      optional: true,
    };
    props.responseFormat = {
      label: "Response Format",
      description: "**Optional**. The format of the response. The default is `json`.",
      type: "string",
      default: "json",
      optional: true,
      options: constants.TRANSCRIPTION_FORMATS,
    };
    props.temperature = common.props.temperature;

    return props;
  },
  methods: {
    createForm({
      file, outputDir,
    }) {
      const form = new FormData();
      form.append("model", "whisper-1");
      if (this.prompt) form.append("prompt", this.prompt);
      if (this.temperature) form.append("temperature", this.temperature);
      if (this.language) form.append("language", this.language);
      if (this.responseFormat) form.append("response_format", this.responseFormat);
      const readStream = fs.createReadStream(join(outputDir, file));
      form.append("file", readStream);
      return form;
    },
    async splitLargeChunks(files, outputDir) {
      for (const file of files) {
        if (fs.statSync(`${outputDir}/${file}`).size / (1024 * 1024) > CHUNK_SIZE_MB) {
          await this.chunkFile({
            file: `${outputDir}/${file}`,
            outputDir,
            index: file.slice(6, 9),
          });
          await execAsync(`rm -f "${outputDir}/${file}"`);
        }
      }
    },
    async chunkFileAndTranscribe({
      file, $,
    }) {
      const outputDir = join("/tmp", "chunks");
      await execAsync(`mkdir -p "${outputDir}"`);
      await execAsync(`rm -f "${outputDir}/*"`);

      await this.chunkFile({
        file,
        outputDir,
      });

      let files = await fs.promises.readdir(outputDir);
      // ffmpeg will sometimes return chunks larger than the allowed size,
      // so we need to identify large chunks and break them down further
      await this.splitLargeChunks(files, outputDir);
      files = await fs.promises.readdir(outputDir);

      return this.transcribeFiles({
        files,
        outputDir,
        $,
      });
    },
    async chunkFile({
      file, outputDir, index,
    }) {
      const ffmpegPath = ffmpegInstaller.path;
      const ext = extname(file);

      const fileSizeInMB = fs.statSync(file).size / (1024 * 1024);
      // We're limited to 26MB per request. Because of how ffmpeg splits files,
      // we need to be conservative in the number of chunks we create
      const conservativeChunkSizeMB = CHUNK_SIZE_MB;
      const numberOfChunks = !index
        ? Math.ceil(fileSizeInMB / conservativeChunkSizeMB)
        : 2;

      if (numberOfChunks === 1) {
        await execAsync(`cp "${file}" "${outputDir}/chunk-000${ext}"`);
        return;
      }

      const { stdout } = await execAsync(`${ffmpegPath} -i "${file}" 2>&1 | grep "Duration"`);
      const duration = stdout.match(/\d{2}:\d{2}:\d{2}\.\d{2}/s)[0];
      const [
        hours,
        minutes,
        seconds,
      ] = duration.split(":").map(parseFloat);

      const totalSeconds = (hours * 60 * 60) + (minutes * 60) + seconds;
      const segmentTime = Math.ceil(totalSeconds / numberOfChunks);

      const command = `${ffmpegPath} -i "${file}" -f segment -segment_time ${segmentTime} -c copy "${outputDir}/chunk-${index
        ? `${index}-`
        : ""}%03d${ext}"`;
      await execAsync(command);
    },
    transcribeFiles({
      files, outputDir, $,
    }) {
      const limiter = new Bottleneck({
        maxConcurrent: 1,
        minTime: 1000 / 59,
      });

      return Promise.all(files.map((file) => {
        return limiter.schedule(() => this.transcribe({
          file,
          outputDir,
          $,
        }));
      }));
    },
    transcribe({
      file, outputDir, $,
    }) {
      const form = this.createForm({
        file,
        outputDir,
      });
      return this.openai.createTranscription({
        $,
        form,
      });
    },
    getFullText(transcriptions = []) {
      return transcriptions.map((t) => t.text || t).join(" ");
    },
  },
  async run({ $ }) {
    const {
      url,
      path,
    } = this;

    if (!url && !path) {
      throw new ConfigurationError("Must specify either File URL or File Path");
    }

    let file;

    if (path) {
      if (!fs.existsSync(path)) {
        throw new ConfigurationError(`${path} does not exist`);
      }

      file = path;
    } else if (url) {
      const ext = extname(url).split("?")[0];

      const response = await axios({
        method: "GET",
        url,
        responseType: "stream",
        timeout: 250000,
      });

      const bufferStream = new stream.PassThrough();
      response.data.pipe(bufferStream);

      const downloadPath = join("/tmp", `audio${ext}`);
      const writeStream = fs.createWriteStream(downloadPath);

      await pipelineAsync(bufferStream, writeStream);

      file = downloadPath;
    }

    const transcriptions = await this.chunkFileAndTranscribe({
      file,
      $,
    });

    if (transcriptions.length) {
      $.export("$summary", "Successfully created transcription");
    }

    return {
      transcription: this.getFullText(transcriptions),
      transcriptions,
    };
  },
};

Action Configuration#

This component may be configured based on the props defined in the component code. Pipedream automatically prompts for input values in the UI.

Label	Prop	Type	Description
OpenAI (ChatGPT)	`openai`	`app`	This component uses the OpenAI (ChatGPT) app.
Audio Upload Type	`uploadType`	`string`	Select a value from the drop down menu:`FileURL`
Language	`language`	`string`	Select a value from the drop down menu:`{ "label": "Afar", "value": "aa" }{ "label": "Abkhaz", "value": "ab" }{ "label": "Avestan", "value": "ae" }{ "label": "Afrikaans", "value": "af" }{ "label": "Akan", "value": "ak" }{ "label": "Amharic", "value": "am" }{ "label": "Aragonese", "value": "an" }{ "label": "Arabic", "value": "ar" }{ "label": "Arabic (Algeria)", "value": "ar-dz" }{ "label": "Arabic (Bahrain)", "value": "ar-bh" }{ "label": "Arabic (Egypt)", "value": "ar-eg" }{ "label": "Arabic (Iraq)", "value": "ar-iq" }{ "label": "Arabic (Jordan)", "value": "ar-jo" }{ "label": "Arabic (Kuwait)", "value": "ar-kw" }{ "label": "Arabic (Lebanon)", "value": "ar-lb" }{ "label": "Arabic (Libya)", "value": "ar-ly" }{ "label": "Arabic (Morocco)", "value": "ar-ma" }{ "label": "Arabic (Oman)", "value": "ar-om" }{ "label": "Arabic (Qatar)", "value": "ar-qa" }{ "label": "Arabic (Saudi Arabia)", "value": "ar-sa" }{ "label": "Arabic (Syria)", "value": "ar-sy" }{ "label": "Arabic (Tunisia)", "value": "ar-tn" }{ "label": "Arabic (U.A.E.)", "value": "ar-ae" }{ "label": "Arabic (Yemen)", "value": "ar-ye" }{ "label": "Assamese", "value": "as" }{ "label": "Avaric", "value": "av" }{ "label": "Aymara", "value": "ay" }{ "label": "Azerbaijani", "value": "az" }{ "label": "Bashkir", "value": "ba" }{ "label": "Belarusian", "value": "be" }{ "label": "Bulgarian", "value": "bg" }{ "label": "Bislama", "value": "bi" }{ "label": "Bambara", "value": "bm" }{ "label": "Bengali", "value": "bn" }{ "label": "Tibetan", "value": "bo" }{ "label": "Breton", "value": "br" }{ "label": "Bosnian", "value": "bs" }{ "label": "Catalan", "value": "ca" }{ "label": "Chechen", "value": "ce" }{ "label": "Chamorro", "value": "ch" }{ "label": "Corsican", "value": "co" }{ "label": "Cree", "value": "cr" }{ "label": "Czech", "value": "cs" }{ "label": "Old Church Slavonic", "value": "cu" }{ "label": "Chuvash", "value": "cv" }{ "label": "Welsh", "value": "cy" }{ "label": "Danish", "value": "da" }{ "label": "German", "value": "de" }{ "label": "Divehi", "value": "dv" }{ "label": "Dzongkha", "value": "dz" }{ "label": "Ewe", "value": "ee" }{ "label": "Greek", "value": "el" }{ "label": "English", "value": "en" }{ "label": "Esperanto", "value": "eo" }{ "label": "Spanish", "value": "es" }{ "label": "Estonian", "value": "et" }{ "label": "Basque", "value": "eu" }{ "label": "Persian", "value": "fa" }{ "label": "Fula", "value": "ff" }{ "label": "Finnish", "value": "fi" }{ "label": "Fijian", "value": "fj" }{ "label": "Faroese", "value": "fo" }{ "label": "French", "value": "fr" }{ "label": "Western Frisian", "value": "fy" }{ "label": "Irish", "value": "ga" }{ "label": "Scottish Gaelic", "value": "gd" }{ "label": "Galician", "value": "gl" }{ "label": "Guaraní", "value": "gn" }{ "label": "Gujarati", "value": "gu" }{ "label": "Manx", "value": "gv" }{ "label": "Hausa", "value": "ha" }{ "label": "Hebrew", "value": "he" }{ "label": "Hindi", "value": "hi" }{ "label": "Hiri Motu", "value": "ho" }{ "label": "Croatian", "value": "hr" }{ "label": "Haitian", "value": "ht" }{ "label": "Hungarian", "value": "hu" }{ "label": "Armenian", "value": "hy" }{ "label": "Herero", "value": "hz" }{ "label": "Interlingua", "value": "ia" }{ "label": "Indonesian", "value": "id" }{ "label": "Interlingue", "value": "ie" }{ "label": "Igbo", "value": "ig" }{ "label": "Nuosu", "value": "ii" }{ "label": "Inupiaq", "value": "ik" }{ "label": "Ido", "value": "io" }{ "label": "Icelandic", "value": "is" }{ "label": "Italian", "value": "it" }{ "label": "Inuktitut", "value": "iu" }{ "label": "Japanese", "value": "ja" }{ "label": "Javanese", "value": "jv" }{ "label": "Georgian", "value": "ka" }{ "label": "Kongo", "value": "kg" }{ "label": "Kikuyu", "value": "ki" }{ "label": "Kwanyama", "value": "kj" }{ "label": "Kazakh", "value": "kk" }{ "label": "Kalaallisut", "value": "kl" }{ "label": "Khmer", "value": "km" }{ "label": "Kannada", "value": "kn" }{ "label": "Korean", "value": "ko" }{ "label": "Kanuri", "value": "kr" }{ "label": "Kashmiri", "value": "ks" }{ "label": "Kurdish", "value": "ku" }{ "label": "Komi", "value": "kv" }{ "label": "Cornish", "value": "kw" }{ "label": "Kyrgyz", "value": "ky" }{ "label": "Latin", "value": "la" }{ "label": "Luxembourgish", "value": "lb" }{ "label": "Ganda", "value": "lg" }{ "label": "Limburgish", "value": "li" }{ "label": "Lingala", "value": "ln" }{ "label": "Lao", "value": "lo" }{ "label": "Lithuanian", "value": "lt" }{ "label": "Luba-Katanga", "value": "lu" }{ "label": "Latvian", "value": "lv" }{ "label": "Malagasy", "value": "mg" }{ "label": "Marshallese", "value": "mh" }{ "label": "Māori", "value": "mi" }{ "label": "Macedonian", "value": "mk" }{ "label": "Malayalam", "value": "ml" }{ "label": "Mongolian", "value": "mn" }{ "label": "Marathi", "value": "mr" }{ "label": "Malay", "value": "ms" }{ "label": "Maltese", "value": "mt" }{ "label": "Burmese", "value": "my" }{ "label": "Nauru", "value": "na" }{ "label": "Norwegian Bokmål", "value": "nb" }{ "label": "Northern Ndebele", "value": "nd" }{ "label": "Nepali", "value": "ne" }{ "label": "Ndonga", "value": "ng" }{ "label": "Dutch", "value": "nl" }{ "label": "Norwegian Nynorsk", "value": "nn" }{ "label": "Norwegian", "value": "no" }{ "label": "Southern Ndebele", "value": "nr" }{ "label": "Navajo", "value": "nv" }{ "label": "Chichewa", "value": "ny" }{ "label": "Occitan", "value": "oc" }{ "label": "Ojibwe", "value": "oj" }{ "label": "Oromo", "value": "om" }{ "label": "Oriya", "value": "or" }{ "label": "Ossetian", "value": "os" }{ "label": "Panjabi", "value": "pa" }{ "label": "Pāli", "value": "pi" }{ "label": "Polish", "value": "pl" }{ "label": "Pashto", "value": "ps" }{ "label": "Portuguese", "value": "pt" }{ "label": "Quechua", "value": "qu" }{ "label": "Romansh", "value": "rm" }{ "label": "Kirundi", "value": "rn" }{ "label": "Romanian", "value": "ro" }{ "label": "Russian", "value": "ru" }{ "label": "Kinyarwanda", "value": "rw" }{ "label": "Sanskrit", "value": "sa" }{ "label": "Sardinian", "value": "sc" }{ "label": "Sindhi", "value": "sd" }{ "label": "Northern Sami", "value": "se" }{ "label": "Sango", "value": "sg" }{ "label": "Sinhala", "value": "si" }{ "label": "Slovak", "value": "sk" }{ "label": "Slovenian", "value": "sl" }{ "label": "Samoan", "value": "sm" }{ "label": "Shona", "value": "sn" }{ "label": "Somali", "value": "so" }{ "label": "Albanian", "value": "sq" }{ "label": "Serbian", "value": "sr" }{ "label": "Swati", "value": "ss" }{ "label": "Southern Sotho", "value": "st" }{ "label": "Sundanese", "value": "su" }{ "label": "Swedish", "value": "sv" }{ "label": "Swahili", "value": "sw" }{ "label": "Tamil", "value": "ta" }{ "label": "Telugu", "value": "te" }{ "label": "Tajik", "value": "tg" }{ "label": "Thai", "value": "th" }{ "label": "Tigrinya", "value": "ti" }{ "label": "Turkmen", "value": "tk" }{ "label": "Tagalog", "value": "tl" }{ "label": "Tswana", "value": "tn" }{ "label": "Tonga", "value": "to" }{ "label": "Turkish", "value": "tr" }{ "label": "Tsonga", "value": "ts" }{ "label": "Tatar", "value": "tt" }{ "label": "Twi", "value": "tw" }{ "label": "Tahitian", "value": "ty" }{ "label": "Uyghur", "value": "ug" }{ "label": "Ukrainian", "value": "uk" }{ "label": "Urdu", "value": "ur" }{ "label": "Uzbek", "value": "uz" }{ "label": "Venda", "value": "ve" }{ "label": "Vietnamese", "value": "vi" }{ "label": "Volapük", "value": "vo" }{ "label": "Walloon", "value": "wa" }{ "label": "Wolof", "value": "wo" }{ "label": "Xhosa", "value": "xh" }{ "label": "Yiddish", "value": "yi" }{ "label": "Yoruba", "value": "yo" }{ "label": "Zhuang", "value": "za" }{ "label": "Chinese", "value": "zh" }{ "label": "Zulu", "value": "zu" }`

Create Transcription (Whisper) with OpenAI (ChatGPT) API on New Inbound Message from SMSTools API

Pipedream makes it easy to connect APIs for OpenAI (ChatGPT), SMSTools and 2,400+ other apps remarkably fast.

Trusted by 1,000,000+ developers from startups to Fortune 500 companies

Developers ♥ Pipedream

1-24of2,400+apps by most popular

1
-
24
of
2,400+
apps by most popular