Text to Speech REST

1
# Install the Deepgram JS SDK
2
# https://github.com/deepgram/deepgram-js-sdk
3
4
npm install @deepgram/sdk

1
import { createClient } from "@deepgram/sdk";
2
import fs from "fs";
3
4
const deepgram = createClient("DEEPGRAM_API_KEY");

1
const text = "Hello, how can I help you today?";
2
3
const getAudio = async () => {
4
  // STEP 1: Make a request and configure the request with options (such as model choice, audio configuration, etc.)
5
  const response = await deepgram.speak.request(
6
    { text },
7
    {
8
      model: "aura-asteria-en",
9
      encoding: "linear16",
10
      container: "wav",
11
    }
12
  );
13
14
  // STEP 2: Get the audio stream and headers from the response
15
  const stream = await response.getStream();
16
  const headers = await response.getHeaders();
17
  if (stream) {
18
    // STEP 3: Convert the stream to an audio buffer
19
    const buffer = await getAudioBuffer(stream);
20
    // STEP 4: Write the audio buffer to a file
21
    fs.writeFile("output.wav", buffer, (err) => {
22
      if (err) {
23
        console.error("Error writing audio to file:", err);
24
      } else {
25
        console.log("Audio file written to output.wav");
26
      }
27
    });
28
  } else {
29
    console.error("Error generating audio:", stream);
30
  }
31
32
  if (headers) {
33
    console.log("Headers:", headers);
34
  }
35
};
36
37
// Helper function to convert the stream to an audio buffer
38
const getAudioBuffer = async (response) => {
39
  const reader = response.getReader();
40
  const chunks = [];
41
42
  while (true) {
43
    const { done, value } = await reader.read();
44
    if (done) break;
45
46
    chunks.push(value);
47
  }
48
49
  const dataArray = chunks.reduce(
50
    (acc, chunk) => Uint8Array.from([...acc, ...chunk]),
51
    new Uint8Array(0)
52
  );
53
54
  return Buffer.from(dataArray.buffer);
55
};
56
57
getAudio();

1
const DEEPGRAM_API_KEY = 'YOUR_DEEPGRAM_API_KEY';
2
const deepgram = createClient(DEEPGRAM_API_KEY);
3
4
const text = "Hello, how can I help you today? My name is Emily and I'm very glad to meet you. What do you think of this new text-to-speech API?";
5
const audioFilePath = 'output.wav'; // Path to save the audio file
6
7
const getAudio = async () => {
8
  const response = await deepgram.speak.request(
9
    { text },
10
    {
11
      model: 'aura-asteria-en',
12
      encoding: 'linear16',
13
      container: 'wav',
14
    }
15
  );
16
17
  const stream = await response.getStream();
18
  if (stream) {
19
    const buffer = await getAudioBuffer(stream);
20
    fs.writeFile(audioFilePath, buffer, (err) => {
21
      if (err) {
22
        console.error('Error writing audio to file:', err);
23
      } else {
24
        console.log('Audio file written to', audioFilePath);
25
      }
26
    });
27
  } else {
28
    console.error('Error generating audio:', stream);
29
  }
30
};
31
32
const getAudioBuffer = async (response) => {
33
  const reader = response.getReader();
34
  const chunks = [];
35
36
  while (true) {
37
    const { done, value } = await reader.read();
38
    if (done) break;
39
40
    chunks.push(value);
41
  }
42
43
  const dataArray = chunks.reduce(
44
    (acc, chunk) => Uint8Array.from([...acc, ...chunk]),
45
    new Uint8Array(0)
46
  );
47
48
  return Buffer.from(dataArray.buffer);
49
};
50
51
getAudio();

1
import fs from 'fs';
2
import { createClient } from '@deepgram/sdk';
3
4
const DEEPGRAM_API_KEY = 'YOUR_DEEPGRAM_API_KEY';
5
const deepgram = createClient(DEEPGRAM_API_KEY);
6
7
const inputText = "Your long text goes here...";
8
9
function segmentTextBySentence(text) {
10
  return text.match(/[^.!?]+[.!?]/g).map((sentence) => sentence.trim());
11
}
12
13
async function synthesizeAudio(text) {
14
  const response = await deepgram.speak.request(
15
    { text },
16
    {
17
      model: 'aura-helios-en',
18
      encoding: 'linear16',
19
      container: 'wav',
20
    }
21
  );
22
23
  const stream = await response.getStream();
24
  if (stream) {
25
    const buffer = await getAudioBuffer(stream);
26
    return buffer;
27
  } else {
28
    throw new Error('Error generating audio');
29
  }
30
}
31
32
const getAudioBuffer = async (response) => {
33
  const reader = response.getReader();
34
  const chunks = [];
35
36
  while (true) {
37
    const { done, value } = await reader.read();
38
    if (done) break;
39
40
    chunks.push(value);
41
  }
42
43
  const dataArray = chunks.reduce(
44
    (acc, chunk) => Uint8Array.from([...acc, ...chunk]),
45
    new Uint8Array(0)
46
  );
47
48
  return Buffer.from(dataArray.buffer);
49
};
50
51
async function main() {
52
  const segments = segmentTextBySentence(inputText);
53
54
  // Create or truncate the output file
55
  const outputFile = fs.createWriteStream("output.mp3");
56
57
  for (const segment of segments) {
58
    try {
59
      const audioData = await synthesizeAudio(segment);
60
      outputFile.write(audioData);
61
      console.log("Audio stream finished for segment:", segment);
62
    } catch (error) {
63
      console.error("Error synthesizing audio:", error);
64
    }
65
  }
66
67
  console.log("Audio file creation completed.");
68
}
69
70
main();