Audio embeddings

This guide shows how you can create audio embeddings using the Marengo video understanding model. For a list of available versions, complete specifications and input requirements for each version, see the Marengo page.

The Marengo video understanding model generates embeddings for all modalities in the same latent space. This shared space enables any-to-any searches across different types of content.

For details on how your usage is measured and billed, see the Pricing page.

Key concepts

This section explains the key concepts and terminology used in this guide:

Asset: Your uploaded content. Once created, you can reference the same asset across multiple operations without uploading the file again.
Embedding: Vector representation of your content.
Embedding task: An asynchronous operation for processing your content and creating embeddings. Contains a status and the resulting embeddings when complete.

Workflow

This guide shows how to upload your audio file as an asset and create embeddings asynchronously. You can also pass a URL or base64-encoded data inline instead of creating an asset; both are shown as commented-out lines in the code examples.

For audio files under 10 minutes, synchronous processing returns embeddings immediately without polling. For details, see the Short audio files (synchronous) section.

Customize your embeddings

You can configure embedding types (audio, transcription), output format (separate, fused, or both), scope (clip or asset), and fixed-duration segmentation.

Use these embeddings for similarity search, content classification, clustering, recommendations, or Retrieval-Augmented Generation (RAG).

Prerequisites

To use the platform, you need an API key:

1
If you don’t have an account, sign up for a free account.
2
Go to the API Keys page.
3
If you need to create a new key, select the Create API Key button. Enter a name and set the expiration period. The default is 12 months.
4
Select the Copy icon next to your key to copy it to your clipboard.

Depending on the programming language you are using, install the TwelveLabs SDK by entering one of the following commands:

$ pip install twelvelabs

Your audio files must meet the following requirements:
- Upload limits: Public audio URLs up to 4 GB or local audio files up to 200 MB.
- Embedding method: Audio files up to 4 hours. This guide uses the asynchronous method. For audio files under 10 minutes, see the synchronous approach below.
- Model capabilities: See the complete requirements for supported formats and specifications.

Complete example

Copy and paste the code below, replacing the placeholders surrounded by <> with your values.

1 import time
2 from twelvelabs import (
3     TwelveLabs,
4     AudioInputRequest,
5     MediaSource,
6     AudioSegmentation,
7     AudioSegmentationFixed
8 )
9 # 1. Initialize the client
10 client = TwelveLabs(api_key="<YOUR_API_KEY>")
11 
12 # 2. Upload an audio file
13 asset = client.assets.create(
14     method="url",
15     url="<YOUR_AUDIO_URL>" # Use direct links to raw media files.
16     # Or use method="direct" and file=open("<PATH_TO_AUDIO_FILE>"", "rb") to upload a local file up to 200 MB
17 )
18 print(f'Created asset: id={asset.id}')
19 
20 # 3. Check the status of the asset
21 print("Waiting for asset to be ready...")
22 while True:
23     asset = client.assets.retrieve(asset.id)
24     if asset.status == "ready":
25         print("Asset is ready")
26         break
27     if asset.status == "failed":
28         raise RuntimeError(f"Asset processing failed: id={asset.id}")
29     time.sleep(5)
30 
31 # 4. Create audio embeddings
32 task = client.embed.v_2.tasks.create(
33     input_type="audio",
34     model_name="marengo3.0",
35     audio=AudioInputRequest(
36         media_source=MediaSource(
37             asset_id=asset.id,
38             # url="<YOUR_AUDIO_URL>", # Use direct links to raw media files
39             # base_64_string="<BASE_64_ENCODED_DATA>",
40         ),
41         # start_sec=0,
42         # end_sec=60,
43         # segmentation=AudioSegmentation(
44         #     fixed=AudioSegmentationFixed(
45         #         duration_sec=6
46         #     )
47         # ),
48         # embedding_option=["audio", "transcription"],
49         # embedding_scope=["clip", "asset"],
50         # embedding_type=["separate_embedding", "fused_embedding"],
51     ),
52 )
53 print(f"Task ID: {task.id}")
54 
55 # 5. Poll until the task is ready
56 while True:
57     task = client.embed.v_2.tasks.retrieve(task_id=task.id)
58 
59     if task.status == "ready":
60         print(f"Task completed")
61         break
62     elif task.status == "failed":
63         print("Task failed")
64         break
65     else:
66         print("Task still processing...")
67         time.sleep(5)
68 
69 
70 # 6. Process the results
71 print(f"\n{'='*80}")
72 print(f"EMBEDDINGS SUMMARY: {len(task.data)} total embeddings")
73 print(f"{'='*80}\n")
74 
75 for idx, embedding_data in enumerate(task.data, 1):
76     print(f"[{idx}/{len(task.data)}] {embedding_data.embedding_option.upper()} | {embedding_data.embedding_scope.upper()}")
77     print(f"├─ Time range: {embedding_data.start_sec}s - {embedding_data.end_sec}s")
78     print(f"├─ Dimensions: {len(embedding_data.embedding)}")
79     print(f"└─ First 10 values: {embedding_data.embedding[:10]}")
80     print()

Code explanation

Python

Node.js

Import the SDK and initialize the client

Create a client instance to interact with the TwelveLabs Video Understanding Platform.
Function call: You call the constructor of the TwelveLabs class.
Parameters:

api_key: The API key to authenticate your requests to the platform.

Return value: An object of type TwelveLabs configured for making API calls.

Upload an audio file

Upload an audio file to create an asset.
Function call: You call the assets.create function.
Parameters:

method: The upload method for your asset. Use url for a publicly accessible or direct to upload a local file. This example uses url.
url or file: The publicly accessible URL of your audio file or an opened file object in binary read mode. This example uses url.

Return value: An object of type Asset. This object contains, among other information, a field named id representing the unique identifier of your asset.

Check the status of the asset

You only need this step for URL uploads larger than 200 MB. The platform processes these files asynchronously.
Function call: You call the assets.retrieve function.
Parameters:

asset_id: The unique identifier of your asset.

Return value: An object of type Asset containing, among other information, a field named status representing the current status of the asset. Check this field until its value is ready.

Create audio embeddings

Create an embedding task to start processing your audio. This operation is asynchronous.
Function call: You call the embed.v_2.tasks.create function.
Parameters:

input_type: The type of content. Set this parameter to audio.
model_name: The model you want to use. This example uses marengo3.0.
audio: An object containing the following properties:
- media_source: An object specifying the source of the audio file. You can specify one of the following:
  - asset_id: The unique identifier of an asset from a previous upload.
  - url: The publicly accessible URL of the audio file.
  - base_64_string: The base64-encoded audio data.
    
    This example uses the asset ID from the previous step.
- (Optional) start_sec: The start time in seconds for processing the audio file. By default, the platform processes audio from the beginning.
- (Optional) end_sec: The end time in seconds for processing the audio file. By default, the platform processes audio to the end of the audio file.
- (Optional) embedding_option: The types of embeddings to generate. Valid values are the following:
  - audio: Generates embeddings for non-verbal audio (musical tones, beeping, environmental sounds).
  - transcription: Generates embeddings for transcribed speech (the actual words spoken in the video).
  You can specify multiple values to generate different types of embeddings. The default value is ["audio", "transcription"].
- (Optional) embedding_scope: The scope for which to generate embeddings. Valid values are the following:
  - clip: Generates one embedding for each segment.
  - asset: Generates one embedding for the entire audio file.
  You can specify multiple scopes to generate embeddings at different levels. The default value is ["clip", "asset"].
- (Optional) segmentation: An object that specifies how the platform divides the audio into segments. Use AudioSegmentation with strategy set to "fixed" and a fixed property containing a duration_sec field to specify the exact duration in seconds for each segment.
- (Optional) embedding_type: An array specifying how to structure the embedding. Use this parameter only when embedding_option specifies two or more values. Valid values are the following:
  - separate_embedding: Returns separate embeddings for each modality specified in embedding_option.
  - fused_embedding: Returns a single combined embedding that integrates all modalities into one vector.
  To receive both types in the same response, set this to ["separate_embedding", "fused_embedding"].

Return value: An object of type TasksCreateResponse containing, among other information, a field named id, which represents the unique identifier of your embedding task. You can use this identifier to track the status of your embedding task.

Monitor the status

The platform requires some time to process audio. Poll the status of the embedding task until processing completes. This example uses a loop to check the status every 5 seconds.
Function call: You repeatedly call the embed.v_2.tasks.retrieve function until the task completes.

Parameters:

task_id: The unique identifier of your embedding task.

Return value: An object of type EmbeddingTaskResponse containing, among other information, the following fields:

status: The current status of the task. The possible values are:
- processing: The platform is creating the embeddings.
- ready: Processing is complete. Embeddings are available in the data field.
- failed: The task failed.
data: When the status is ready, this field contains a list of embedding objects. Each embedding object includes:
- embedding: The embedding vector (a list of floats).
- embedding_option: The type of embedding. Possible values are audio, transcription, and fused. The platform returns fused only when embedding_type includes fused_embedding.
- embedding_scope: The scope of the embedding (clip or asset).
- start_sec: The start time of the segment in seconds.
- end_sec: The end time of the segment in seconds.

Process the results

This example iterates through the embeddings in the data field and prints the embedding type, scope, time range, dimensions, and the first 10 vector values for each segment.

Short audio files (synchronous)

For audio files shorter than 10 minutes, you can use a synchronous approach that returns embeddings immediately without requiring polling.

1 import time
2 from twelvelabs import (
3     TwelveLabs,
4     AudioInputRequest,
5     MediaSource,
6     # AudioSegmentation,
7     # AudioSegmentationFixed
8 )
9 
10 # 1. Initialize the client
11 client = TwelveLabs(api_key="<YOUR_API_KEY>")
12 
13 # 2. Upload an audio file
14 asset = client.assets.create(
15     method="url",
16     url="<YOUR_AUDIO_URL>", # Use direct links to raw media files
17     # Or use method="direct" and file=open("<PATH_TO_AUDIO_FILE>", "rb") to upload a file from the local file system
18 )
19 print(f'Created asset: id={asset.id}')
20 
21 # 3. Check the status of the asset
22 print("Waiting for asset to be ready...")
23 while True:
24     asset = client.assets.retrieve(asset.id)
25     if asset.status == "ready":
26         print("Asset is ready")
27         break
28     if asset.status == "failed":
29         raise RuntimeError(f"Asset processing failed: id={asset.id}")
30     time.sleep(5)
31 
32 # 4. Create audio embeddings
33 response = client.embed.v_2.create(
34     input_type="audio",
35     model_name="marengo3.0",
36     audio=AudioInputRequest(
37         media_source=MediaSource(
38             asset_id=asset.id,
39             # url="<YOUR_AUDIO_URL>", # Use direct links to raw media files
40             # base_64_string="<BASE_64_ENCODED_DATA>",
41         ),
42         # start_sec=0,
43         # end_sec=60,
44         # segmentation=AudioSegmentation(
45         #     fixed=AudioSegmentationFixed(
46         #         duration_sec=6
47         #     )
48         # ),
49         # embedding_option=["audio", "transcription"],
50         # embedding_scope=["clip", "asset"],
51         # embedding_type=["separate_embedding", "fused_embedding"],
52     ),
53 )
54 
55 # 5. Process the results
56 print(f"\n{'='*80}")
57 print(f"EMBEDDINGS SUMMARY: {len(response.data)} total embeddings")
58 print(f"{'='*80}\n")
59 
60 for idx, embedding_data in enumerate(response.data, 1):
61     print(f"[{idx}/{len(response.data)}] {embedding_data.embedding_option.upper()} | {embedding_data.embedding_scope.upper()}")
62     print(f"├─ Time range: {embedding_data.start_sec}s - {embedding_data.end_sec}s")
63     print(f"├─ Dimensions: {len(embedding_data.embedding)}")
64     print(f"└─ First 10 values: {embedding_data.embedding[:10]}")
65     print()

All the fields of the audio object function similarly to the asynchronous approach.