diff --git a/src/main.cpp b/src/main.cpp
index 2eab1c2..622ea70 100644
--- a/src/main.cpp
+++ b/src/main.cpp
@@ -1,6 +1,7 @@
 #include <iostream>
 #include <boost/property_tree/ptree.hpp>
 #include <boost/property_tree/xml_parser.hpp>
+#include <boost/optional.hpp>
 #include <format.h>
 #include <tclap/CmdLine.h>
 #include "audioInput/WaveFileReader.h"
@@ -69,6 +70,14 @@ ptree createXmlTree(const path& filePath, const map<centiseconds, Phone>& phones
 	return tree;
 }
 
+// Tell TCLAP how to handle boost::optional
+namespace TCLAP {
+	template<>
+	struct ArgTraits<boost::optional<string>> {
+		typedef TCLAP::StringLike ValueCategory;
+	};
+}
+
 int main(int argc, char *argv[]) {
 	// Define command-line parameters
 	const char argumentValueSeparator = ' ';
@@ -76,6 +85,7 @@ int main(int argc, char *argv[]) {
 	cmd.setExceptionHandling(false);
 	cmd.setOutput(new NiceCmdLineOutput());
 	TCLAP::UnlabeledValueArg<string> inputFileName("inputFile", "The input file. Must be a sound file in WAVE format.", true, "", "string", cmd);
+	TCLAP::ValueArg<boost::optional<string>> dialog("d", "dialog", "The text of the dialog.", false, boost::optional<string>(), "string", cmd);
 
 	try {
 		// Parse command line
@@ -88,7 +98,10 @@ int main(int argc, char *argv[]) {
 		map<centiseconds, Phone> phones;
 		{
 			ProgressBar progressBar;
-			phones = detectPhones([&inputFileName]() { return createAudioStream(inputFileName.getValue()); }, progressBar);
+			phones = detectPhones(
+				[&inputFileName]() { return createAudioStream(inputFileName.getValue()); },
+				dialog.getValue(),
+				progressBar);
 		}
 		std::cerr << "Done" << std::endl;
 
diff --git a/src/phoneExtraction.cpp b/src/phoneExtraction.cpp
index bf2bf70..e8ccfa9 100644
--- a/src/phoneExtraction.cpp
+++ b/src/phoneExtraction.cpp
@@ -7,6 +7,8 @@
 #include "platformTools.h"
 #include "tools.h"
 #include <format.h>
+#include <s3types.h>
+#include <regex>
 
 extern "C" {
 #include <pocketsphinx.h>
@@ -17,6 +19,7 @@ extern "C" {
 }
 
 using std::runtime_error;
+using std::invalid_argument;
 using std::unique_ptr;
 using std::shared_ptr;
 using std::string;
@@ -24,6 +27,8 @@ using std::vector;
 using std::map;
 using boost::filesystem::path;
 using std::function;
+using std::regex;
+using std::regex_replace;
 
 unique_ptr<AudioStream> to16kHzMono(unique_ptr<AudioStream> stream) {
 	// Downmix, if required
@@ -33,7 +38,7 @@ unique_ptr<AudioStream> to16kHzMono(unique_ptr<AudioStream> stream) {
 
 	// Downsample, if required
 	if (stream->getFrameRate() < 16000) {
-		throw runtime_error("Audio sample rate must not be below 16kHz.");
+		throw invalid_argument("Audio sample rate must not be below 16kHz.");
 	}
 	if (stream->getFrameRate() != 16000) {
 		stream.reset(new SampleRateConverter(std::move(stream), 16000));
@@ -133,7 +138,7 @@ void sphinxErrorCallback(void* user_data, err_lvl_t errorLevel, const char* form
 	*errorString += message;
 }
 
-vector<s3wid_t> recognizeWords(unique_ptr<AudioStream> audioStream, ps_decoder_t& recognizer, ProgressSink& progressSink) {
+vector<string> recognizeWords(unique_ptr<AudioStream> audioStream, ps_decoder_t& recognizer, ProgressSink& progressSink) {
 	// Convert audio stream to the exact format PocketSphinx requires
 	audioStream = to16kHzMono(std::move(audioStream));
 
@@ -153,19 +158,55 @@ vector<s3wid_t> recognizeWords(unique_ptr<AudioStream> audioStream, ps_decoder_t
 	if (error) throw runtime_error("Error ending utterance processing for word recognition.");
 
 	// Collect words
-	vector<s3wid_t> result;
+	vector<string> result;
 	int32_t score;
 	for (ps_seg_t* it = ps_seg_iter(&recognizer, &score); it; it = ps_seg_next(it)) {
-		// Get word
 		const char* word = ps_seg_word(it);
-		s3wid_t wordId = dict_wordid(recognizer.dict, word);
-
-		result.push_back(wordId);
+		result.push_back(word);
 	}
 
 	return result;
 }
 
+// Splits dialog into words, doing minimal preprocessing.
+// A robust solution should use TTS logic to cope with numbers, abbreviations, unknown words etc.
+vector<string> extractDialogWords(string dialog) {
+	// Convert to lower case
+	boost::algorithm::to_lower(dialog);
+
+	// Insert silences where appropriate
+	dialog = regex_replace(dialog, regex("[,;.:!?] "), " <sil> ");
+
+	// Remove all undesired characters
+	dialog = regex_replace(dialog, regex("[^a-z.'\\-0-9<>]"), " ");
+
+	// Collapse whitespace
+	dialog = regex_replace(dialog, regex("\\s+"), " ");
+
+	// Trim
+	boost::algorithm::trim(dialog);
+
+	// Ugly hack: Remove trailing period
+	if (boost::algorithm::ends_with(dialog, ".")) {
+		dialog.pop_back();
+	}
+
+	// Split into words
+	vector<string> result;
+	boost::algorithm::split(result, dialog, boost::is_space());
+	return result;
+}
+
+vector<s3wid_t> getWordIds(const vector<string>& words, dict_t& dictionary) {
+	vector<s3wid_t> result;
+	for (const string& word : words) {
+		s3wid_t wordId = dict_wordid(&dictionary, word.c_str());
+		if (wordId == BAD_S3WID) throw invalid_argument(fmt::format("Unknown word '{}'.", word));
+		result.push_back(wordId);
+	}
+	return result;
+}
+
 map<centiseconds, Phone> getPhoneAlignment(const vector<s3wid_t>& wordIds, unique_ptr<AudioStream> audioStream, ps_decoder_t& recognizer, ProgressSink& progressSink) {
 	// Create alignment list
 	lambda_unique_ptr<ps_alignment_t> alignment(
@@ -236,7 +277,11 @@ map<centiseconds, Phone> getPhoneAlignment(const vector<s3wid_t>& wordIds, uniqu
 	return result;
 }
 
-map<centiseconds, Phone> detectPhones(std::function<std::unique_ptr<AudioStream>(void)> createAudioStream, ProgressSink& progressSink) {
+map<centiseconds, Phone> detectPhones(
+	std::function<std::unique_ptr<AudioStream>(void)> createAudioStream,
+	boost::optional<std::string> dialog,
+	ProgressSink& progressSink)
+{
 	// Discard Pocketsphinx output
 	err_set_logfp(nullptr);
 
@@ -253,11 +298,16 @@ map<centiseconds, Phone> detectPhones(std::function<std::unique_ptr<AudioStream>
 		auto recognizer = createSpeechRecognizer(*config.get());
 
 		ProgressMerger progressMerger(progressSink);
-		ProgressSink& wordRecognitionProgressSink = progressMerger.addSink(1.0);
+		ProgressSink& wordRecognitionProgressSink = progressMerger.addSink(dialog ? 0.0 : 1.0);
 		ProgressSink& alignmentProgressSink = progressMerger.addSink(0.5);
 
-		// Recognize words
-		vector<s3wid_t> wordIds = recognizeWords(createAudioStream(), *recognizer.get(), wordRecognitionProgressSink);
+		// Get words
+		vector<string> words = dialog
+			? extractDialogWords(*dialog)
+			: recognizeWords(createAudioStream(), *recognizer.get(), wordRecognitionProgressSink);
+
+		// Look up words in dictionary
+		vector<s3wid_t> wordIds = getWordIds(words, *recognizer->dict);
 
 		// Align the word's phones with speech
 		map<centiseconds, Phone> result = getPhoneAlignment(wordIds, createAudioStream(), *recognizer.get(), alignmentProgressSink);
diff --git a/src/phoneExtraction.h b/src/phoneExtraction.h
index c2fee5f..a01f3b2 100644
--- a/src/phoneExtraction.h
+++ b/src/phoneExtraction.h
@@ -7,5 +7,9 @@
 #include "Phone.h"
 #include "centiseconds.h"
 #include "progressBar.h"
+#include <boost/optional/optional.hpp>
 
-std::map<centiseconds, Phone> detectPhones(std::function<std::unique_ptr<AudioStream>(void)> createAudioStream, ProgressSink& progressSink);
+std::map<centiseconds, Phone> detectPhones(
+	std::function<std::unique_ptr<AudioStream>(void)> createAudioStream,
+	boost::optional<std::string> dialog,
+	ProgressSink& progressSink);