{ "cells": [ { "cell_type": "code", "execution_count": 1, "id": "ab6c7d18", "metadata": {}, "outputs": [ { "name": "stdout", "output_type": "stream", "text": [ "Populating the interactive namespace from numpy and matplotlib\n" ] } ], "source": [ "%pylab inline" ] }, { "cell_type": "code", "execution_count": 2, "id": "8a4404a3", "metadata": {}, "outputs": [], "source": [ "import os\n", "os.environ[\"CUDA_VISIBLE_DEVICES\"] = \"0\"" ] }, { "cell_type": "code", "execution_count": 3, "id": "e4f3f4a2", "metadata": {}, "outputs": [], "source": [ "import json\n", "from suno_utils.audio import Audio\n", "from suno_utils.tasks.asr import transcribe, LmConfig\n", "\n", "DATA_DIR = \"/home/georg/notebooks/customers/nooks/sample_10/\"\n", "\n", "with open(DATA_DIR + \"annotations.json\") as f:\n", " annotations = json.load(f)\n", " \n", "filepaths = [DATA_DIR + fn for fn in os.listdir(DATA_DIR) if fn.endswith(\".wav\")]" ] }, { "cell_type": "code", "execution_count": 4, "id": "938b015c", "metadata": {}, "outputs": [], "source": [ "# voicemail:\n", "# 1 channel (prob use asr on both even if 2)\n", "# terms: \"your call\", \"voicemail\", \"voice mail\", \"messages\", \"mailbox\", \"after the tone\", \"your message\", \"at the tone\"\n", "# \"hi you have reached XXX XXX i can't make it to the phone right now, but please leave me a message\"\n", "# (one person talking for at least 5 mins)\n", " \n", "# human:\n", "# only 1 channel of conversation\n", " \n", "# dialtree: (leading up to a call?)\n", "# 1 channel (prob use asr on both even if 2)\n", "# terms: \"thank you for calling\", \"please listen carefully as our menu options have changed\", \"please press [0-9]\"" ] }, { "cell_type": "code", "execution_count": 18, "id": "86ada964", "metadata": {}, "outputs": [], "source": [ "voicemail_search_terms = [\n", " \"your call\", \n", " \"voicemail\",\n", " \"voice mail\",\n", " \"message\",\n", " \"messages\", \n", " \"mailbox\", \n", " \"after the tone\", \n", " \"at the tone\", \n", " \"you have reached\", \n", " \"you've reached\",\n", " \"you reached\",\n", "]\n", "dialtree_search_terms = [\n", " \"thank you for calling\", \n", " \"please listen carefully\", \n", " \"our menu options have changed\",\n", " \"our menu options changed\",\n", " \"please press\",\n", "]\n", "voicemail_re = re.compile(r\"|\".join([\n", " r\"\\b\" + re.escape(s) + r\"\\b\" for s in sorted(voicemail_search_terms, key=lambda x: len(x), reverse=True)\n", "]))\n", "dialtree_re = re.compile(r\"|\".join([\n", " r\"\\b\" + re.escape(s) + r\"\\b\" for s in sorted(dialtree_search_terms, key=lambda x: len(x), reverse=True)\n", "]))\n", "hotwords = []\n", "for s in voicemail_search_terms + dialtree_search_terms:\n", " hotwords.extend(s.split())\n", "hotwords = set(hotwords)" ] }, { "cell_type": "code", "execution_count": 19, "id": "58d8303b", "metadata": {}, "outputs": [ { "data": { "application/vnd.jupyter.widget-view+json": { "model_id": "6e5deffb05734daebb1ff3d165ef01ef", "version_major": 2, "version_minor": 0 }, "text/plain": [ "Transcribing: 0%| | 0/3 [00:00 0 and ts[1] <= intro_duration_s\n", " ]) / intro_duration_s * 100, 1)\n", " keyword_duration_s = 20\n", " text = \"\".join([t for t, ts, _ in transcript if ts[1] <= keyword_duration_s])\n", " voicemail_terms = voicemail_re.findall(text)\n", " dialtree_terms = dialtree_re.findall(text)\n", " print(\"{} -- ({})\".format(uid, label))\n", " human_pct = (1 - (np.clip(speech_pct, a_min=5, a_max=30) - 5) / 25)\n", " print(\"is_human: {}%\".format(round(human_pct * 100, 1)))\n", " if human_pct >= 0.7 and len(voicemail_terms) == 0 and len(dialtree_terms) == 0:\n", " voicemail_pct = 0\n", " dialtree_pct = 0\n", " elif len(voicemail_terms) == 0 and len(dialtree_terms) == 0:\n", " voicemail_pct = 0.5\n", " dialtree_pct = 0.5\n", " else:\n", " voicemail_pct = len(voicemail_terms) / (len(voicemail_terms) + len(dialtree_terms))\n", " dialtree_pct = len(dialtree_terms) / (len(voicemail_terms) + len(dialtree_terms))\n", " print(\"is_voicemail: {}%\".format(round(voicemail_pct * 100, 1)))\n", " print(\"is_dialtree:: {}%\".format(round(dialtree_pct * 100, 1)))\n", " print(\"-\" * 10)" ] }, { "cell_type": "code", "execution_count": null, "id": "90818b55", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "eda295ee", "metadata": {}, "outputs": [], "source": [] }, { "cell_type": "code", "execution_count": null, "id": "7cb49441", "metadata": {}, "outputs": [], "source": [] } ], "metadata": { "kernelspec": { "display_name": "Python 3 (ipykernel)", "language": "python", "name": "python3" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 3 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", "version": "3.8.10" } }, "nbformat": 4, "nbformat_minor": 5 }