holm/whisper.cpp
1
1
Fork 0
forked from extern/whisper.cpp
Code Issues Pull Requests Packages Projects Releases Wiki Activity

talk : improve prompting

Browse Source
This commit is contained in:
Georgi Gerganov
2022-12-12 23:44:36 +02:00
parent 930c693989
commit a613f16aec
4 changed files with 39 additions and 79 deletions
Download Patch File Download Diff File Expand all files Collapse all files

101
examples/talk/talk.cpp
View File

@ -473,56 +473,15 @@ std::string transcribe(whisper_context * ctx, const whisper_params & params, con
return result;
}
// compute similarity between two strings using Levenshtein distance
float similarity(const std::string & s0, const std::string & s1) {
const size_t len0 = s0.size() + 1;
const size_t len1 = s1.size() + 1;
const std::string k_prompt =
R"(This is a dialogue between {0} (A) and a person (B). The dialogue so far is:
std::vector<int> col(len1, 0);
std::vector<int> prevCol(len1, 0);
B: Hello {0}, how are you?
A: I'm fine, thank you.
{1}
Here is how {0} (A) continues the dialogue:
for (size_t i = 0; i < len1; i++) {
prevCol[i] = i;
}
for (size_t i = 0; i < len0; i++) {
col[0] = i;
for (size_t j = 1; j < len1; j++) {
col[j] = std::min(std::min(1 + col[j - 1], 1 + prevCol[j]), prevCol[j - 1] + (s0[i - 1] == s1[j - 1] ? 0 : 1));
}
col.swap(prevCol);
}
const float dist = prevCol[len1 - 1];
return 1.0f - (dist / std::max(s0.size(), s1.size()));
}
// generated with ChatGPT
std::map<std::string, std::string> k_prompts = {
{ "Santa",
R"(Kid: Hi Santa! Are you real?
Santa: Of course I am, my dear! Ho ho ho!
Kid: Can you please bring me a new toy for Christmas?
Santa: I'll see what I can do, but you have to make sure to be a good boy or girl and listen to your parents.
Kid: I will, Santa! Thank you!
Santa: You're welcome, little one. Merry Christmas! Ho ho ho!
Kid: Can you tell me how you deliver all the presents to all the kids in the world in one night?
Santa: It's a secret, but I have a lot of help from my elves and my magical sleigh. And I have a special route that I follow to make sure I visit every child.
Kid: Wow, that's amazing! Can I please have a ride in your sleigh sometime?
Santa: I'm sorry, but only good boys and girls get to ride in my sleigh.
)" },
{ "Kid",
R"(Kid: Hi Santa! Are you real?
Santa: Of course I am, my dear! Ho ho ho!
Kid: Can you please bring me a new toy for Christmas?
Santa: I'll see what I can do, but you have to make sure to be a good boy or girl and listen to your parents.
Kid: I will, Santa! Thank you!
Kid: Can you tell me how you deliver all the presents to all the kids in the world in one night?
Santa: It's a secret, but I have a lot of help from my elves and my magical sleigh. And I have a special route that I follow to make sure I visit every child.
Kid: Wow, that's amazing! Can I please have a ride in your sleigh sometime?
)" },
};
A:)";
int main(int argc, char ** argv) {
whisper_params params;
@ -579,7 +538,7 @@ int main(int argc, char ** argv) {
int n_iter = 0;
bool is_running = true;
bool force_speak = params.person == "Kid";
bool force_speak = false;
float prob0 = 0.0f;
float prob = 0.0f;
@ -587,19 +546,13 @@ int main(int argc, char ** argv) {
std::vector<float> pcmf32_cur;
std::vector<float> pcmf32_prompt;
if (k_prompts.find(params.person) == k_prompts.end()) {
fprintf(stderr, "%s: unknown person '%s'\n", __func__, params.person.c_str());
return 1;
}
gpt2_set_prompt(ctx_gpt, "");
gpt2_set_prompt(ctx_gpt, k_prompts.at(params.person).c_str());
const int voice_id = rand()%6;
const std::string person_other = params.person == "Santa" ? "Kid" : "Santa";
const int voice_id = params.person == "Santa" ? 5 : 2;
fprintf(stderr, "gpt-2: prompt_base:\n");
fprintf(stderr, "gpt-2: prompt:\n");
fprintf(stderr, "========================\n\n");
fprintf(stderr, "%s\n", gpt2_get_prompt(ctx_gpt));
fprintf(stderr, "%s\n", ::replace(k_prompt, "{0}", params.person).c_str());
fprintf(stderr, "========================\n\n");
// main loop
@ -636,13 +589,12 @@ int main(int argc, char ** argv) {
audio.get(params.voice_ms, pcmf32_cur);
std::string text_heard = "Hey little one, what do you want for Christmas?";
std::string text_heard = "";
if (!force_speak) {
text_heard = ::trim(::transcribe(ctx_wsp, params, pcmf32_cur, prob0, t_ms));
}
force_speak = false;
// remove text between brackets using regex
{
std::regex re("\\[.*?\\]");
@ -667,13 +619,15 @@ int main(int argc, char ** argv) {
const std::vector<gpt_vocab::id> tokens = gpt2_tokenize(ctx_gpt, text_heard.c_str());
if (text_heard.empty() || tokens.empty()) {
if (text_heard.empty() || tokens.empty() || force_speak) {
fprintf(stdout, "%s: Heard nothing, skipping ...\n", __func__);
audio.clear();
continue;
}
force_speak = false;
fprintf(stdout, "%s: Heard '%s%s%s', (t = %d ms)\n", __func__, "\033[1m", text_heard.c_str(), "\033[0m", (int) t_ms);
std::string prompt_base = gpt2_get_prompt(ctx_gpt);
@ -681,9 +635,11 @@ int main(int argc, char ** argv) {
std::string text_to_speak;
{
text_heard = person_other + ": " + text_heard;
prompt_base += "B: " + text_heard + "\n";
text_to_speak = gpt2_gen_text(ctx_gpt, (prompt_base + text_heard + "\n").c_str(), params.max_tokens);
std::string prompt = ::replace(::replace(k_prompt, "{0}", params.person), "{1}", prompt_base);
text_to_speak = gpt2_gen_text(ctx_gpt, prompt.c_str(), params.max_tokens);
text_to_speak = std::regex_replace(text_to_speak, std::regex("[^a-zA-Z0-9\\.,\\?!\\s\\:\\'\\-]"), "");
text_to_speak = text_to_speak.substr(0, text_to_speak.find_first_of("\n"));
@ -703,13 +659,20 @@ int main(int argc, char ** argv) {
}
}
prompt_base += text_heard + "\n" + text_to_speak + "\n";
prompt_base += "A:" + text_to_speak + "\n";
{
prompt = ::replace(::replace(k_prompt, "{0}", params.person), "{1}", prompt_base);
printf("===============\n");
printf("prompt:\n");
printf("%s\n", prompt.c_str());
printf("===============\n");
}
}
printf("%s\n", text_to_speak.c_str());
//printf("========================\n");
//printf("gpt-2: prompt_base:\n'%s'\n", prompt_base.c_str());
//printf("gpt-2: prompt_base:\n%s\n", prompt_base.c_str());
//printf("========================\n");
gpt2_set_prompt(ctx_gpt, prompt_base.c_str());