core-transcription/speaker-diarization-with-async-chunking.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "<a href=\"https://www.assemblyai.com/\">\n",
    "<img width=\"300\"\n",
    "src=\"data:image/png;base64,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\">\n",
    "</a>\n",
    "</img>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Speaker Diarization with Async Chunking\n",
    "\n",
    "This guide uses AssemblyAI and [Nvidia's NeMo](https://catalog.ngc.nvidia.com/orgs/nvidia/teams/nemo/models/titanet_large) framework. We'll be using TitaNet, a state of the art open source model that is trained for speaker recognition tasks. TitaNet will allow us to generate audio embeddings for speakers, which can be used to identify semantic similarity matches between two speakers.\n",
    "\n",
    "## Get Started\n",
    "\n",
    "Before we begin, make sure you have an AssemblyAI account and an API key. You can [sign up for an AssemblyAI account](https://www.assemblyai.com/app) and get your API key from your dashboard.\n",
    "\n",
    "## Install Dependencies"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "%pip install pytorch\n",
    "%pip install nemo_toolkit['all']\n",
    "%pip install ffmpeg\n",
    "%pip install assemblyai"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## AssemblyAI Setup, Transcript Setup, and Load the Model Using NeMO\n",
    "In this section, we'll import dependencies and add functions to transcribe and store transcript IDs if needed."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "import assemblyai as aai\n",
    "import requests\n",
    "import json\n",
    "import time\n",
    "import requests\n",
    "import copy\n",
    "from pydub import AudioSegment\n",
    "import os\n",
    "import nemo.collections.asr as nemo_asr\n",
    "\n",
    "speaker_model = nemo_asr.models.EncDecSpeakerLabelModel.from_pretrained(\"nvidia/speakerverification_en_titanet_large\")\n",
    "\n",
    "assemblyai_key = \"YOUR_API_KEY\"\n",
    "\n",
    "headers = {\n",
    "    \"authorization\": assemblyai_key\n",
    "}"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Helper Functions\n",
    "\n",
    "The function below requests a transcript based on a transcript ID."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_transcript(transcript_id):\n",
    "    polling_endpoint = f\"https://api.assemblyai.com/v2/transcript/{transcript_id}\"\n",
    "\n",
    "    while True:\n",
    "        transcription_result = requests.get(polling_endpoint, headers=headers).json()\n",
    "\n",
    "        if transcription_result['status'] == 'completed':\n",
    "            # print(\"Transcript ID:\", transcript_id)\n",
    "            return(transcription_result)\n",
    "            break\n",
    "\n",
    "        elif transcription_result['status'] == 'error':\n",
    "            raise RuntimeError(f\"Transcription failed: {transcription_result['error']}\")\n",
    "\n",
    "        else:\n",
    "            time.sleep(3)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Our main inference function will make use of these functions:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def download_wav(presigned_url, output_filename):\n",
    "    # Download the WAV file from the presigned URL\n",
    "    response = requests.get(presigned_url)\n",
    "    if response.status_code == 200:\n",
    "        print(\"downloading...\")\n",
    "        with open(output_filename, 'wb') as f:\n",
    "            f.write(response.content)\n",
    "        print(\"successfully downloaded file:\", output_filename)\n",
    "    else:\n",
    "        raise Exception(\"Failed to download file, status code: {}\".format(response.status_code))\n",
    "\n",
    "# Function to identify the longest monologue of each speaker from each clip\n",
    "# you pass in the utterances and it returns the longest monologue from each speaker on that file\n",
    "def find_longest_monologues(utterances):\n",
    "    longest_monologues = {}\n",
    "    current_monologue = {}\n",
    "    last_speaker = None  # Track the last speaker to identify interruptions\n",
    "\n",
    "    for utterance in utterances:\n",
    "        speaker = utterance['speaker']\n",
    "        start_time = utterance['start']\n",
    "        end_time = utterance['end']\n",
    "\n",
    "        if speaker not in current_monologue:\n",
    "            current_monologue[speaker] = {\"start\": start_time, \"end\": end_time}\n",
    "            longest_monologues[speaker] = []\n",
    "        else:\n",
    "            # Extend monologue only if it's the same speaker speaking continuously\n",
    "            if current_monologue[speaker][\"end\"] == start_time and last_speaker == speaker:\n",
    "                current_monologue[speaker][\"end\"] = end_time\n",
    "            else:\n",
    "                monologue_length = current_monologue[speaker][\"end\"] - current_monologue[speaker][\"start\"]\n",
    "                new_entry = (monologue_length, copy.deepcopy(current_monologue[speaker]))\n",
    "\n",
    "                if len(longest_monologues[speaker]) < 1 or monologue_length > min(longest_monologues[speaker], key=lambda x: x[0])[0]:\n",
    "                    if len(longest_monologues[speaker]) == 1:\n",
    "                        longest_monologues[speaker].remove(min(longest_monologues[speaker], key=lambda x: x[0]))\n",
    "\n",
    "                    longest_monologues[speaker].append(new_entry)\n",
    "\n",
    "                current_monologue[speaker] = {\"start\": start_time, \"end\": end_time}\n",
    "\n",
    "        last_speaker = speaker  # Update the last speaker\n",
    "\n",
    "    # Check the last monologue for each speaker\n",
    "    for speaker, monologue in current_monologue.items():\n",
    "        monologue_length = monologue[\"end\"] - monologue[\"start\"]\n",
    "        new_entry = (monologue_length, monologue)\n",
    "        if len(longest_monologues[speaker]) < 1 or monologue_length > min(longest_monologues[speaker], key=lambda x: x[0])[0]:\n",
    "            if len(longest_monologues[speaker]) == 1:\n",
    "                longest_monologues[speaker].remove(min(longest_monologues[speaker], key=lambda x: x[0]))\n",
    "            longest_monologues[speaker].append(new_entry)\n",
    "\n",
    "    return longest_monologues\n",
    "\n",
    "# Create clips of each long monologue and embed the clip\n",
    "# you pass in the file path and the longest monologue objects returned by the find_longest_monologues function.\n",
    "# This function will create new audio file clips which contain only the longest monologue from each speaker\n",
    "def clip_and_store_utterances(audio_file, longest_monologues):\n",
    "    # Load the full conversation audio\n",
    "    full_audio = AudioSegment.from_wav(audio_file)\n",
    "    full_audio = full_audio.set_channels(1)\n",
    "\n",
    "    utterance_clips = []\n",
    "\n",
    "    for speaker, monologues in longest_monologues.items():\n",
    "        for _, monologue in monologues:\n",
    "            start_ms = monologue['start']\n",
    "            end_ms = monologue['end']\n",
    "            clip = full_audio[start_ms:end_ms]\n",
    "            clip_filename = f\"{speaker}_monologue_{start_ms}_{end_ms}.wav\"\n",
    "            clip.export(clip_filename, format=\"wav\")\n",
    "\n",
    "            utterance_clips.append({\n",
    "                'clip_filename': clip_filename,\n",
    "                'start': start_ms,\n",
    "                'end': end_ms,\n",
    "                'speaker': speaker\n",
    "            })\n",
    "\n",
    "    print(\"Total Number of Monologue Clips Found: \", len(utterance_clips))\n",
    "\n",
    "    return utterance_clips\n",
    "\n",
    "# This function uses NeMO to compare two files\n",
    "def compare_embeddings(utterance_clip, reference_file):\n",
    "    verification_result = speaker_model.verify_speakers(\n",
    "        utterance_clip,\n",
    "        reference_file\n",
    "    )\n",
    "    return verification_result"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Inference\n",
    "\n",
    "Add the links to the WAV file clips you have stored on your server."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "file_one = \"YOUR_FILE_1\"\n",
    "file_two = \"YOUR_FILE_2\"\n",
    "file_three = \"YOUR_FILE_3\"\n",
    "\n",
    "download_wav(file_one, \"testone.wav\")\n",
    "download_wav(file_two, \"testtwo.wav\")\n",
    "download_wav(file_three, \"testthree.wav\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from pydub import AudioSegment\n",
    "\n",
    "# Store utterances from each clip, keyed by clip index\n",
    "clip_utterances = {}\n",
    "\n",
    "# Dictionary to track known speaker identities across all clips\n",
    "# Maps current clip speaker labels to a unified speaker label\n",
    "speaker_identity_map = {}\n",
    "\n",
    "def process_clips(clip_transcript_ids, audio_files):\n",
    "    global clip_utterances, speaker_identity_map\n",
    "\n",
    "    # This will store the longest clip filenames for each speaker from the previous clips\n",
    "    previous_speaker_clips = {}\n",
    "\n",
    "    for clip_index, (transcript_id, audio_file) in enumerate(zip(clip_transcript_ids, audio_files)):\n",
    "        transcript = get_transcript(transcript_id)\n",
    "        utterances = transcript['utterances']\n",
    "        clip_utterances[clip_index] = utterances # Store utterances for the current clip\n",
    "\n",
    "        longest_monologues = find_longest_monologues(utterances)\n",
    "\n",
    "        # Process the longest monologues for clipping and storing\n",
    "        current_speaker_clips = {}\n",
    "        for speaker, monologue_data in longest_monologues.items():\n",
    "            clip_and_store_utterances(audio_file, {speaker: monologue_data})\n",
    "            longest_clip = f\"{speaker}_monologue_{monologue_data[0][1]['start']}_{monologue_data[0][1]['end']}.wav\"\n",
    "            current_speaker_clips[speaker] = longest_clip\n",
    "\n",
    "        if clip_index == 0:\n",
    "            speaker_identity_map = {speaker: speaker for speaker in longest_monologues.keys()}\n",
    "            previous_speaker_clips = current_speaker_clips.copy()\n",
    "        else:\n",
    "            # Compare all new speakers against all base speakers from previous clips\n",
    "            for new_speaker, new_clip in current_speaker_clips.items():\n",
    "                for base_speaker, base_clip in previous_speaker_clips.items():\n",
    "                    if compare_embeddings(new_clip, base_clip):\n",
    "                        speaker_identity_map[new_speaker] = base_speaker\n",
    "                        break\n",
    "                else:\n",
    "                    # If no match is found, assign a new label\n",
    "                    new_label = chr(ord(max(speaker_identity_map.values(), key=lambda x: ord(x))) + 1)\n",
    "                    speaker_identity_map[new_speaker] = new_label\n",
    "\n",
    "            # Update the previous_speaker_clips for the next iteration\n",
    "            previous_speaker_clips.update(current_speaker_clips)\n",
    "\n",
    "        # Update utterances with the new speaker labels for the current clip\n",
    "        for utterance in clip_utterances[clip_index]:\n",
    "            original_speaker = utterance['speaker']\n",
    "            # Update only if there's a change in speaker identity\n",
    "            if original_speaker in speaker_identity_map:\n",
    "                utterance['speaker'] = speaker_identity_map[original_speaker]\n",
    "\n",
    "\n",
    "# Add your clip transcript IDs\n",
    "clip_transcript_ids = [\n",
    "    \"YOUR_TRANSCRIPT_ID_1\",\n",
    "    \"YOUR_TRANSCRIPT_ID_2\",\n",
    "    \"YOUR_TRANSCRIPT_ID_3\"\n",
    "]\n",
    "\n",
    "# Add filepaths to your downloaded files\n",
    "audio_files = [\n",
    "    \"/testone.wav\",\n",
    "    \"/testtwo.wav\",\n",
    "    \"/testthree.wav\"\n",
    "]\n",
    "\n",
    "process_clips(clip_transcript_ids, audio_files)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## New Clip Utterances\n",
    "\n",
    "The clip utterances returned by the `process_clips` function will contain the corrected utterances, which can be seen by printing out the utterances or by using the `display_transcript` function."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "print(clip_utterances)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "def display_transcript(transcript_data):\n",
    "    for clip_index, utterances in transcript_data.items():\n",
    "        print(f\"Clip {clip_index + 1}:\")\n",
    "        for utterance in utterances:\n",
    "            speaker = utterance['speaker']\n",
    "            text = utterance['text']\n",
    "            print(f\"  Speaker {speaker}: {text}\")\n",
    "        print(\"\\n\")  # Add an extra newline for spacing between \n",
    "        \n",
    "        \n",
    "display_transcript(clip_utterances)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Additional Resources\n",
    "\n",
    "- [Async chunking](https://github.com/AssemblyAI-Solutions/async-chunking)\n",
    "- [AsyncChunkPy: Near-Realtime Python Speech-to-Text App](https://github.com/AssemblyAI-Solutions/async-chunk-py)\n",
    "- [Guide for Identifying speakers across multiple podcasts with AssemblyAI and TitaNet](https://docs.google.com/document/d/1xdOvY1LM2lGUNRZCf3kNQLSs5OCUfo74_MHDGm_jYc0/edit?tab=t.0#heading=h.59f375jz72wm)\n",
    "- [Multi-Speaker Voice Identification and Diarization with AssemblyAI, Pinecone, and Nvidia's TitaNet Model](https://colab.research.google.com/drive/1vqpYcPLEjDjMJ9WvP8C1gvOIflBHs0VS?usp=sharing)"
   ]
  }
 ],
 "metadata": {
  "language_info": {
   "name": "python"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}