semantically-searchable transcription

Multi-Module Pipeline: Semantically-Searchable Transcription

This document details a multi-modular pipeline that takes in an audio file, transcribes it, and makes the result semantically (vector) searchable.

Semantic search involves an understanding of the intent and context behind natural language queries to deliver more relevant and flexible results. This pipeline can help in enhancing video content management, improving the accessibility of spoken information, and enabling advanced analytics on audio, among other possibilities.

The document is divided into the following sections:

Pipeline Setup
Processing an Input File
Performing Semantic Search

Pipeline Setup

To achieve what we've described above, let's set up a pipeline sequentially consisting of the following modules:

A transcribe module.
A json-to-txt module.
A parser module.
A text-embedder module.
A vector-db module.

We use the json-to-txt and parser combination, which combines the transcribed snippets into one document and then splices it again, to make sure that any pauses in speech don't make for partial snippets that can confuse the text-embedder model.

Pipeline setup is accomplished through the create_pipeline method, as follows:

# create a pipeline as detailed above
pipeline = krixik.create_pipeline(
    name="multi_semantically_searchable_transcription", module_chain=["transcribe", "json-to-txt", "parser", "text-embedder", "vector-db"]
)

Processing an Input File

A pipeline's valid input formats are determined by its first module—in this case, a transcribe module. Therefore, this pipeline only accepts audio file inputs.

Lets take a quick look at a test file before processing.

# examine contents of input file
import IPython

IPython.display.Audio(data_dir + "input/Interesting Facts About Colombia.mp3")

We will use the default models for every module in the pipeline, so the modules argument of the process method doesn't need to be explicitly input.

# process the file through the pipeline, as described above
process_output = pipeline.process(
    local_file_path=data_dir + "input/Interesting Facts About Colombia.mp3",  # the initial local filepath where the input file is stored
    local_save_directory=data_dir + "output",  # the local directory that the output file will be saved to
    expire_time=60 * 30,  # process data will be deleted from the Krixik system in 30 minutes
    wait_for_process=True,  # wait for process to complete before returning IDE control to user
    verbose=False,
)  # do not display process update printouts upon running code

The output of this process is printed below. To learn more about each component of the output, review documentation for the process method.

Because the output of this particular module-model pair is a FAISS database file, the process output is null. However, the output file has been saved to the location noted in the process_output_files key. The file_id of the processed input is used as a filename prefix for the output file.

# nicely print the output of this process
print(json.dumps(process_output, indent=2))

{
  "status_code": 200,
  "pipeline": "multi_semantically_searchable_transcription",
  "request_id": "0efa11da-ef8c-4b94-9595-05960d937f96",
  "file_id": "9177ce5e-1662-4b5d-bf67-7b538a0d2837",
  "message": "SUCCESS - output fetched for file_id 9177ce5e-1662-4b5d-bf67-7b538a0d2837.Output saved to location(s) listed in process_output_files.",
  "warnings": [],
  "process_output": null,
  "process_output_files": [
    "../../../data/output/9177ce5e-1662-4b5d-bf67-7b538a0d2837.faiss"
  ]
}

Performing Semantic Search

Krixik's semantic_search method enables semantic (a.k.a. vector) search on documents processed through certain pipelines. Given that the semantic_search method both embeds the query and performs the search, it can only be used with pipelines containing both a text-embedder module and a vector-db module in immediate succession.

Since our pipeline satisfies this condition, it has access to the semantic_search method. Let's use it to query our text with natural language, as shown below:

# perform semantic_search over the file in the pipeline
semantic_output = pipeline.semantic_search(query="Let's talk about the country of Colombia", file_ids=[process_output["file_id"]])

# nicely print the output of this process
print(json.dumps(semantic_output, indent=2))

{
  "status_code": 200,
  "request_id": "c6c1a446-0bfb-459f-82a2-8051257b3eb1",
  "message": "Successfully queried 1 user file.",
  "warnings": [],
  "items": [
    {
      "file_id": "9177ce5e-1662-4b5d-bf67-7b538a0d2837",
      "file_metadata": {
        "file_name": "krixik_generated_file_name_lvwjnpbsbh.mp3",
        "symbolic_directory_path": "/etc",
        "file_tags": [],
        "num_vectors": 13,
        "created_at": "2024-06-05 14:50:47",
        "last_updated": "2024-06-05 14:50:47"
      },
      "search_results": [
        {
          "snippet": "And today we are gonna be looking more at Columbia in our Columbia Part 2 video.",
          "line_numbers": [
            1
          ],
          "distance": 0.267
        },
        {
          "snippet": " That's episode looking at the great country of Columbia.",
          "line_numbers": [
            1
          ],
          "distance": 0.276
        },
        {
          "snippet": "But if you're new here, join me every single Monday to learn about new countries from around the world.",
          "line_numbers": [
            1
          ],
          "distance": 0.32
        },
        {
          "snippet": "Which just reminds me guys, this is part of our Columbia playlist.",
          "line_numbers": [
            1
          ],
          "distance": 0.33
        },
        {
          "snippet": "It's name, a bit of its history, the type of people that live there, land size and all that jazz.",
          "line_numbers": [
            1
          ],
          "distance": 0.345
        }
      ]
    }
  ]
}