The particular dialect or language that a person chooses to use on any occasion is called a code Copyright © 2025 The Author IDEAS is licensed under CC-BY-SA 4.0 License Issued by English study program of IAIN Palopo 5443 IDEAS Journal on Language Teaching and Learning, Linguistics and Literature ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) Volume 13, Number 2, December 2025 pp. 5443 - 5460 Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill Israeni Iskandar1, Nasriandi2, Syahrir3 1Pendidikan Bahasa Inggris, Universitas Muhammadiyah Palopo, Sulawesi Selatan 2,3Ilmu Pendidikan, Universitas Hasyim Asy’ari, Jombang, Jawa Timur Corresponding E-Mail: iskandarisraeni8@gmail.com Received: 2024-07-27 Accepted: 2025-10-19 DOI: 10.24256/ideas. v13i2.5374 Abstract This study evaluates the effectiveness of CAKE application in improving the speaking skills of grade II students of SMA Negeri 6 Palopo. CAKE is an artificial intelligence (AI)- based English learning application with a speech recognition feature that provides immediate feedback on the user's pronunciation. The study used a pseudo quantitative experimental method with data collection techniques in the form of pre-test and post- test. The research sample was selected by simple randomization from the student population. The results showed that there was no statistically significant difference between the experimental class using the CAKE app and the control class using conventional methods based on the results of the independent T-test. However, the app still had a positive impact in improving students' speaking skills. Significant improvements were seen in the “Excellent” and “Moderate” categories in the experimental group compared to the control group. The motivational factor of the students also had an effect on the results of this improvement. Keywords: Artificial Intelligent 1; Speaking Skill 2; language 3 (Alphabetical) Introduction Speaking skills are very important in the current era of globalization because communication plays a key role in success in various fields. Language is used as a means of communication, and English is considered an international language used around the world to communicate with people from different regions, countries and continents.(Berlianti, 2022) Some of these reasons are then taken into consideration to include English into the learning curriculum in Indonesia. Education in Indonesia has included English as one of the foreign languages that must be learned due to the development of the times and technology. Individuals must master four skills in learning English namely http://u.lipi.go.id/1457703302 mailto:iskandarisraeni8@gmail.com Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5444 speaking, listening, reading and writing. These skills are important to achieve the goal of learning English, whether to work abroad, fulfill company requirements, or continue higher education.(Abrori, 2023) Speaking in English is very important because it is one of the basic skills needed. According to Scoot Thornbury, speaking situations reveal the different styles and interpersonal functions of each speaker as well as their written responses. Transactional acts serve as information brokers and service providers, while interpersonal functions build and maintain relationships.(scott thornbury, 2017) The researcher observed that many high school students could not speak English well, even for simple self-introductions. Many students do not know how to answer basic questions in English, which shows their low speaking ability. The average score of students only reached the standard of 60-70, and only a few exceeded this standard.(Noorman Haryadi, 2020) This is due to several factors that are caused by the teaching staff and the students themselves. Some of the obstacles faced in teaching English are students' lack of understanding of English subjects, student motivation, self-confidence for fear of being wrong and the assumption that English is difficult to learn. That's why this affects students' English speaking skills.(I Gede Santika Yasa et al., 2023) The development of speaking skills in English is influenced by teaching methods in the classroom. Many teachers still use old methods that make students feel bored and not excited about learning.(Dhanan Abimanto, 2023) Along with the times, many learning applications are created by experts to make it easier for students to learn, this can be utilized by teachers, especially with the existence of AI (Artificial intelligence), which is a system used by humans to assist in several jobs.(Oyebola Olusola Ayeni et al., 2024)AI makes learning English easier and follows the development of students by making learning videos, voice recording virtue and can be directly assessed by Ai and questions that can be quickly assessed and recognize mistakes, and AI-based applications use more modern, less boring and flexible features used anywhere, this is also included in one application, namely CAKE.(Irma Herawati, 2023) CAKE is an application that uses AI Speech Recognition to provide immediate feedback on the user's pronunciation. This app quickly adapts to the times and keeps updating the way it learns to speak English through song, video or movie snippets. This is different from other apps that are only based on reading or listening to audio.(Iman, 2020) Indriani Febriani's research found that using the CAKE app can improve students' pronunciation. However, this study focused on the use of the app in games, not in learning. Other researchers are interested in seeing if the CAKE app can improve students' speaking skills through teaching methods involving this app.(Febriani et al., 2023) IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5445 English speaking skills are indispensable in the 5.0 era due to the increasingly advanced technological developments. This skill is a big consideration in the world of work. Therefore, it is important for individuals to learn to speak English as an international language.(Noorman Haryadi, 2020). Method This research uses quantitative experimental methods to test the effectiveness of experimental variables. In this study, researchers applied the experimental quantitative method to determine whether the CAKE application could improve students' speaking skills. The data collection techniques used were pre-test, experimentation, and post-test. This study used a quasi-experiment involving two classes, where one class was taught using the CAKE app and the other class used conventional teaching methods. Pre-test and post-test were conducted by giving oral questions to measure students' speaking ability before and after the treatment.(Nasrum et al., 2023) The population of this study was the XI grade students of SMA Negeri 6 Palopo with a total population of 90 students, consisting of 30 students in each class. The research sample was taken by simple randomization, in which each member of the population had the same opportunity to become a sample. The total sample taken was 15 students from each class, with the consideration that the grade XI students had adequate English-speaking ability although not fluent. The sample selection was also based on the suggestion of the English teacher who understood the research plan. Before starting learning, the researcher will carry out a pre-test first so that the researcher can see how the students' abilities before being given a teaching treatment using the AI method after entering and carrying out the treatment 8 times. Where the meeting consists of one first test, namely the pre- test, then teaching six times then at the last meeting for the post-test. For the pre- test and post-test, researchers will give tests in the form of oral questions which are then answered by students orally. After conducting the pre-test, the researcher began teaching in the classroom using a projector connected to a cellphone using an HDMI Dongle: ANYCAST HDMI connected via a wifi network. During the learning process students watch the video after watching, questions will appear that they will answer according to the video that has been watched in the form of conjunction questions, continued words and vocabulary, after that the video will be played once again then the students imitate what is said in the video but the sound is recorded and immediately given an assessment by AI and will be seen the mistake, the teacher will immediately correct and see the results of the next recording whether the pronunciation is good or still the same as before. Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5446 And repeat the word in the app 5 to 6 times until the student's pronunciation is clear and precise. After that they will memorize 5 sentences and 20 vocabulary words from the video in the video and the application includes homework to repeat memorization learning so that students can remember the previous material. This lasted for 6 meetings. In the control class, students used the usual teaching method by the teacher without using the app. The main data collection technique in this study was a test that measured students' speaking skills. The researcher administered a pre-test and post-test consisting of five oral questions tailored to the material that had been compiled. Students' test results were assessed using JB Heaton's assessment rubric which includes three aspects: accuracy, fluency, and comprehension. Accuracy assesses how similar the student's speech is to the native speaker, fluency assesses how fluently the student speaks without stuttering, and comprehension assesses the clarity and understandability of the student's speech. Scoring is done on a scale of 10-90 based on these three aspects.(Heaton, 1988) Numbers Assessment Criteria Score 1 Accuracy - Students do not use accents, or regional dialects or are only slightly carried away by their mother's tongue. only slightly carried over by the mother's tongue and sounds like a native speaker 80-90 - Pronunciation has a slight regional accent and there are few errors in pronunciation. 60-70 - Pronunciation still uses regional accents or mother tongue, grammatical errors. 40-50 - Pronunciation is heavily influenced by mother tongue and grammar. 20-30 - Many mistakes in pronunciation, grammar and there are some words or sentences that cannot be understood. 10 2. Fluency (Smoothness) - Speaks without too much memory, without expression struggles to find vocabulary or sentences and doesn't have much pause time. 80-90 - Sometimes still looking for vocabulary and sentences, speaking fluently although there are some pauses. 60-70 - The student is still looking for vocabulary and sentences, while speaking quite smoothly, sometimes still pauses but manages to convey the general meaning of what he wants to explain. 40-50 - Has a long pause time, often has to look for 20-30 IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5447 meaning, vocabulary or sentences. - Student often pauses his/her speech for a long period of time, often stops while speaking, almost gives up trying and has a very limited vocabulary. 10 3. Comprehensibility (Understanding) - It is easy for the examiner to understand the student's words, when speaking only a little to explain the general meaning of the words. 90 - The student's point is clearly conveyed even if some clarification is needed. 70-80 - Most of what the students say is easy to follow. The intention is clearly conveyed but still requires assistance to help convey the meaning of what is being said. 50-60 - The examiner is able to understand most of what the student says but still lacks a lot of vocabulary. 30-40 - Only a small part can be understood (some short sentences and simple phrases, the examiner can find it difficult to understand what the student is saying. 20 - Almost none of what the student says can be understood, even though the examiner has clarified many times but still cannot understand what the student says. 10 From the table above, you can see the scores obtained by students, after students get scores based on the table criteria, then the final calculation is carried out using SPSS. After the researcher gets the test results, students in this enter the graduation category as follows After the researcher gets the test results, students in this enter the graduation category as follows: No Categories Score Intervals 1 Very good 86-95 2 Good 71-85 3 Average 56-70 4 Poor 41-55 5 Very Poor 10-40 After collecting data and analyzing all and getting test results, the researcher in this calculates the results using SPPS. Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5448 Results This study aims to evaluate the effectiveness of the CAKE application in improving students' speaking skills. Researchers compared the results of the experimental class using the CAKE application and the control group using conventional learning methods. By using the Independent Sample T-test or difference test. The researcher used an independent sample T-test because the two subjects tested were from different classes and different people. Based on the SPSS formula, the Independent Sample T-test is the most suitable for the research conducted. The following are the results of data processing using the independent sample T-test SPSS application. Frequency and percentage of student’s pre-test of experimental and control group the table above presents data on score classification results for the Experimental group and the Control group before the treatment (pre-test). Each score classification is based on a predetermined score range. For the Experimental group, no participant obtained a “Very Good” or “Very Poor” classification. The majority of participants, 40%, fell into the "Average" classification. Followed by 13.33 % of participants classified as "Bad", and 6.66% of participants in the "Good" classification. Meanwhile, in the Control group, a different pattern was seen. No participant obtained a “Very Good” or “Good” classification. The majority of participants, namely 60%, fell into the “Average” classification. Followed by 13% of participants classified as "Poor", and 6.66 % of participants in the "Very Poor" classification. In conclusion, before treatment was administered, both groups had relatively similar score distributions, with the majority of participants falling into the "Average" classification. However, the Experimental group had a higher proportion of “Good” and “Poor” classifications compared to the Control group. In the context of a pre-test comparison before treatment, the Experimental group showed slightly greater variation in the distribution of its scores, with some participants achieving both “Good” and “Poor” classifications, while the majority remained in the “Average” classification. On the other hand, the Control group showed a distribution of scores more centered around the “Average” classification, with little variation in the other classifications. Classification Range of Scores Experiment Control Group N % N % Very Good 86-95 0 0.00 0 0.00 Good 71-85 1 6.66 0 0.00 Average 56-70 6 40.00 9 60.00 Poor 41-55 2 13.33 2 13.00 Very Poor 10-40 0 0.00 1 6.66 Amount 15 100 15 100 IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5449 mean and std. deviation of student’s pretest of experiment and control group The data above shows the pre-test scores before the treatment was given to two different class groups: the control class group and the experimental class group. The average pre-test score for the experimental class group (55.8667) was higher than the control class group (54.0667). This shows that before the treatment was given, the experimental class group had slightly higher scores than the control class group. However, it should be remembered that this difference is not necessarily statistically significant, because in the scores of the two groups there is only a slight difference in scores. Frequency and percentage of students post -test of experimental and control group Classification Range of Scores Experiment Control Group N % N % Very Good 86-95 6 40.00 0 0.00 Good 71-85 1 6.66 4 26.66 Average 56-70 6 40.00 8 53.33 Poor 41-55 2 13.33 1 6.66 Very Poor 10-40 0 0.00 2 13.33 Amount 15 100 15 100 After being given treatment using the CAKE application, there was a change in the distribution of post-test scores between the experimental and control groups. In the experimental group there was a significant increase in the "Very Good" and "Medium" categories with percentages of 40.00 % and 40.00% respectively. Meanwhile, in the control group there was a lower increase in the same category, namely "Good" at 26.66 % and "Medium" at 53.33%. This shows that the treatment given through the CAKE application tends to have a more positive impact on learning achievement, especially in the "Very Good" and 54.0667 12.56677 55.8667 9.91295 0 10 20 30 40 50 60 Mean Std. Deviation control class experiment class Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5450 "Medium" categories compared to the control group. However, this cannot show a statistically significant difference. To find out whether there is a statistically significant difference, you need to look at the results of the normality test, homogeneity test, and the T-test results. 1.6. Normality Test Table In the table above it can be seen that the significant results are > 0.05, which means that if the sig (significant) results for both classes are > 0.05 then the data is considered normal. The table above shows that the Kolmogorov- Smimov and Shapiro-walk values are more than 0.05, so it can be concluded that the data is normal. Test of Homogeneity of Variance Levene Statistics df1 df2 Sig. Student Learning Outcomes Based on Mean ,506 1 28 ,483 Based on Median ,445 1 28 ,510 Based on Median and with adjusted df ,445 1 25,813 ,511 Based on trimmed mean ,516 1 28 ,478 Tests of Normality learning model Kolmogorov-Smirnov a Shapiro-Wilk Statistics df Sig. Statisti cs df Sig. Student Learning Outcomes control class ,206 15 ,086 ,891 15 ,070 experimental class ,235 15 ,025 ,891 15 ,070 a. Lilliefors Significance Correction IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5451 To see whether the data is homogeneous or not, it can be seen from the first table, namely based on the mean, if the significance value is more than 0.05 then the data can be said to be homogeneous. The data above shows that the sig (significant) value based on mean is 0.483, so it can be concluded that the data above is homogeneous data. To make a decision whether the calculation results from SPSS.20 are statistically significant or not, you need to look at the results of the T-test analysis. mean and std. deviation of student’s post-test of experiment and control group The data above shows the post-test scores before the treatment was given to two different class groups: the control class group and the experimental class group. The average pre-test score for the experimental class group (59.6667) was higher than the control class group (70.8). This shows that after the treatment was given, the experimental class group had higher scores than the control class group. However, it is important to remember that this difference is not necessarily statistically significant, because to prove whether the two values are statistically 0.506 1 28 0.4830.445 1 28 0.510.445 1 25.813 0.5110.516 1 28 0.4780 5 10 15 20 25 30 Levene Statistics df1 df2 Sig. Chart Title Based on Mean Based on Median Based on Median and with adjusted df Based on trimmed mean 54.0667 12.56677 55.8667 9.91295 0 10 20 30 40 50 60 Mean Std. Deviation control class experiment class Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5452 significant, we need to look at the results of the sig (2-tailed) analysis in the independent sample T-test table. Bar chart Frequency and percentage of student’s pre-test of experiment and control group Bar chart Frequency and percentage of student’s post-test of experimental and control group 0 10 20 30 40 50 60 70 86-95 71-85 56-70 41-55 Oct-40 Very Good Good Average Poor Very Poor experiment class control Class 0 10 20 30 40 50 60 86-95 71-85 56-70 41-55 Oct-40 Very Good Good Average Poor Very Poor experiment class control class IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5453 Levene's Test for Equality of Variances t-test for Equality of Means F Sig. t df Sig. (2- taile d) Mean Differenc e Std. Erro r Diffe renc e 95% Confidence Interval of the Difference Lower Upp er Student Learning Outcomes Equal variances assumed ,506 ,48 3 - 1,96 6 28 ,059 - 11.13333 5.66 263 - 22.732 71 .466 04 Equal variances not assumed - 1,96 6 27, 99 8 ,059 - 11.13333 5.66 263 - 22.732 75 .466 09 The table above shows the significance value in the results of Levene's test for equality of variances Sig. is> 0.005, then the significance data in this case used is Equal variances assumed. Based on the significance results of Sig. (2-tailed) is 0.059> 0.05, so it can be concluded that there is no statistical difference in the post-test results in the control class and the experimental class on the post-test results. Although the mean of the experimental group is higher than the control group, the difference is not considered statistically significant. So it can be said that cake application is not statistically efficient in improving students' speaking ability. Discussion After calculating the students' grades, the researcher then categorized the calculated results according to the existing category table. After categorizing, you can see the results of the grades from the class before being given treatment, the two groups have a relatively similar score distribution, with the majority of participants falling into the "Average" classification. However, the experimental group had a higher proportion of "Good" classifications of 6.66% and "Bad" of 13.33% compared to the control group. After being given treatment after approximately 6 meetings using the CAKE application. This is supported by Werdamurti's research entitled “Improving Speaking Skills in English Through the Three Step Interview Technique in the SMKNegeri 4 Tanah Grogot Class” said in his research that before the treatment of students' scores were relatively low this was due to students not having enough vocabulary to speak so that what if they were spoken to and nervous to speak they would automatically use limited vocabulary to speak.(Werdamurti, 2023) Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5454 After carrying out the post-test and categorizing the results of the students' scores, there was an increase in the results of the students' scores where in the experimental class there was an increase as much as in the experimental class where the increase in the "Very Good" category was 40.00% while the control class did not. After getting the scores and categorizing them, the researchers then processed the student learning outcomes data using the SPSS software application to see whether the cake application had a statistical effect on improving students' speaking skills. After the results of the data processing carried out by SPSS using the Independent Sample T test, by first carrying out a normality test that the data is normally distributed with a significance of .086 and .025 > 0.05 then the data is normally distributed which can be continued with the homogeneity test which is seen From the first table of the normality table, it shows a significant value of .483 > 0.05, so the data is declared homogeneous. Then, because the results of the homogeneity and normality tests were normal, the Independent Sample T test was then carried out which stated that the results of the statistical calculations were .059 > 0.05, so the results were declared not statistically significant. However, in the bar diagram of frequency and percentage of students, it can be seen that there was an increase, although only 10%, this increase was obtained after the treatment where the researcher entered teaching using the CAKE application, which is the application. Where the researcher directly applied the cake application at the learning stage, this increase was not very significant and was influenced by the students' learning motivation. The results of this study are reinforced by the findings of Ahmad Hasanudin in his journal entitled “Peningkatan Keterampilan Berbicara dengan Metode Debat Plus pada Pembelajaran Bahasa Inggris” said that students' speaking skills improved after the application of the debate ples method, he said that this happened because of the students' motivation because they saw their friends who were smooth in speaking. This can certainly be interpreted that motivation in building the spirit of learning is very necessary to provoke students' learning spirit.(Ahmad Hasanudin, 2021) Research by researcher shows that CAKE as an AI-based application is not significant in improving speaking skills in statistical tests. However, it shows a positive impact in improving students' speaking skills in English. This can be seen in the significant increase in the distribution of post-test scores in the experimental group. This can be seen from the greater improvement in the "Very Good" and "Averange" categories in the experimental group compared to the control group. This is in line with the results of research by Bin Zouh et al in their journal entitled "An Investigation into Artificial Intelligence Speech Evaluation Programs with Automatic Feedback for Developing EFL Learners' Speaking Skills" published in 2023 which stated that there was no statistically significant increase between group A and group B in the analysis test results using analysis of variance (ANOVA). However, the AI speaking speech evaluation program can IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5455 provide more feedback regarding fluency and pronunciation in text reading aloud tasks while the speech content in spontaneous speaking tasks and offers more speaking feedback. intuitive and suitable exercise suggestions. (Zou et al., 2023) The difference between Bin Zouh's research and researchers' research lies in the population size and to the extent that Bin Zou's population and samples are larger and involve 40 subjects compared to researchers who used 30 subjects in total, the next difference can also be seen in the statistical tests where Bin Zouh used analysis of variance (ANOVA) on Microsoft Exel with XLSTAT software while the researcher used the Independent Sample T-test on the SPSS 20 application and the final difference. The researcher focused on the cake application while Bin Zouh on AI-based speech. He researchers' findings show that CAKE as an AI-based application cannot improve speaking skills using statistical tests. However, it shows a positive impact in improving students' speaking skills in English. This can be seen in the significant increase in the distribution of post-test scores in the experimental group. This can be seen from the greater improvement in the "Very Good" and "Medium" categories in the experimental group compared to the control group. This is also reinforced by the findings from research by Hongliang Qiao1Aruna Zhao, Aruna Zhao in her research entitled "Artificial intelligence-based language learning: illuminating the impact on speaking skills and self-regulation in Chinese EFL context" in 2023 said that CET does not has a significant influence on one of the dependent variables. However, AI learners showed greater improvements in speaking and self-regulation skills compared to non-AI learners.(Qiao & Zhao, 2023) These positive results can be attributed to the creative and engaging environment that AI provides for interactive speaking activities. The difference between the researcher's research and Hongliang Qiao1Aruna Zhao's research lies in the AI application used, where Hongliang Qiao1Aruna Zhao's research used CET while the researcher's research used the CAKE application, and in the data processing application the researcher used SPSS while Hongliang Qiao1Aruna Zhao used the Mplus Version 7 application. The cake application is suitable for improving students' speaking skills because the main features in this application are intended to improve students' speaking skills. This is supported by Siti Amalia Rachmawati's research entitled “Pengaruh Penggunaan Aplikasi Cake Apps terhadap Kemampuan Berbicara Siswa“ said in her journal that the cake application is good at improving speaking skills because it is flexible and can be used anywhere independently and also in learning it is very helpful. (Siti Amalia Rachmawati, 2023) The difference between Siti Amalia Rachmawati's research and the researcher's research lies in the number of samples where her research used 20 samples while the researcher's research used 15 samples and her research focused on junior high school while the researcher focused on high school Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5456 students. The results of statistical calculations that are not significant are influenced by student motivation in learning, there are several factors including the first during the learning process there are some students who are less active and not interested in learning this is because students feel that because this application is flexible and can be used anywhere students are more relaxed and more interested in other things such as talking with their friends about games, gossiping, looking for something etc. The second is the English language which is still foreign to the Indonesian tongue which makes it difficult for students to pronounce some sentences. The second is the English language which is still foreign to the Indonesian tongue which makes it difficult for students to pronounce some sentences. This affects the enthusiasm for learning and students' understanding of the learning that is taking place in front of the class, the lack of student motivation in learning is also discussed by Maharani Putri in her journal entitled “Motivasi Belajar Mahasiswa Pada Pemahaman Membaca Bahasa Inggris” saying that motivation greatly affects the way students learn and their level of understanding because what if students have motivation they will be able to absorb and manage the information they get easily. (Maharani Putri, 2023) However, there are also studies that statistically show that the CAKE application is statistically proven to improve students' speaking skills. In a journal researched by Chusnul Chotimah and Sinta Julia Pratiwi entitled The Use of CAKE Application on Students' Speaking Skill in English for Specitific Purposes (EPS) ", said that the CAKE application was statistically proven to improve students' speaking skills. The difference between the researcher's research and this study is that the population taken by the researcher is university students. After reading the results of this study and the researcher's research can conclude that this application is more suitable for use by students where this application can be fully utilized by students who are more aware of the use of the application and they can learn both in class and outside the classroom.(Chotimah, 2022) CAKE application in improving students' speaking skills because in the application there are features that support among others 1. Real-Time Monitoring and Feedback: CAKE application allows teachers or learners to monitor learners' verbal interactions in real-time. Thus, they can provide immediate feedback related to clarity, fluency, and accuracy in language use. 2. Conversation Recording and Analysis: The CAKE app has a feature to record learners' conversations or readings of texts that the app has provided, so that it can analyze learners recorded speech. This analysis can include evaluation of language patterns, common errors, and areas for improvement. This technology is called AI Speech Recognition where the results of the voice recording in this be analyzed as a whole and then the results in this appear on the reading text that the application provides the application in this provide a score from A-E where the designed IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5457 assessment is the accuracy of pronunciation, clarity in speaking and fluency in speaking. Furthermore, Language Recognition and Practice: The app can also provide exercises tailored to the learner's ability level, including the introduction of new vocabulary, sentence patterns, and structured speaking exercises. The app comes with features time where the app in this automatically calculate the number of days and hours the learner has used the app. 4. Customization and Adaptive Learning: CAKE allows personalization of learning based on the individual abilities and needs of the students. This can help in maximizing speaking learning outcomes by focusing on aspects that require more attention. When the user enters the app, there in this be a few questions that in this be asked which include age and the question of how far the individual English learner has come after answering these questions. The system in this automatically display some quizzes and videos and learning according to the level of the application user. This is in line with the opinion of Abdullahi B. Saka, et al. in his journal entitled "Conversational artificial intelligence in the AEC industry: A review of present status, challenges and opportunities" that if an application has several features that have been written above, it is said that the AI is good and can improve students' speaking skills.(Saka et al., 2023) AI (Artificial Intelligence) does make it easier for humans to find information about education, the number of AI-based teaching applications that appear in the world of education makes students helpful in their learning as well as this cake application besides the cake application there are many more AI- based applications that can help improve students' speaking skills, one of which is Duolingo according to Rahmawati et al, in his journal entitled “Improving Speaking Skills By Using Project-Based Learning Whit The Duolingo Application For The Second Grade Students Of Smakon Aceh” that this AI-based application is effective in improving students' English speaking skills for both individuals and groups.(Rahmawati et al, 2024) Conclusion Based on the results of research conducted by researchers, the use of the CAKE application in improving students' English-speaking skills shows several findings. Although there was no statistically significant difference between the experimental class that used the CAKE application in the learning process and the control group in the Independent Sample T-test, this application had a positive impact that was visible from increasing post-test scores in the experimental group, especially in the "Very Good" category. " and "Medium". Features such as real-time monitoring and feedback, conversation recording and analysis, language recognition, and adaptive learning enable this app to provide effective support in speaking learning. However, student motivation factors also influence the results Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5458 of this increase. Based on the researcher's findings, it can be said that the CAKE application can improve students' speaking skills. although the improvement obtained is small, this is due to the lack of student learning motivation and awareness of learning. The shortcomings of this study are that the research time is not long enough and the research was conducted in high school. Suggestions for further research CAKE application may be better used by junior high school students. Suggestions for teachers try to use this application in learning sessions that require students to speak because this application makes it easier for students to learn to speak English. Acknowledgement (if any) The researcher would like to express his gratitude for the support from all parties who participated to help make the research a success. Firstly, the researcher would like to express his praise and thanks for the presence of the most merciful Allah who has facilitated all the researcher's efforts. Secondly, the researcher would like to thank the prayers of his parents. researchers and their support. Next, I would like to thank the research supervisors, examining lecturers and the campus staff who were agile so that all the procedures carried out by the researchers were made easier. and don't forget to also thank the Palopo State Islamic Institute for providing a place for researchers to publish research journals. References Abrori, N. A. (2023). Pentingnya Pembelajaran Bahasa Inggris di Sekolah. Caban Dinas Pendidikan Wilayah I Provensi Jawa Tengah. https://cabdindikwil1.com/blog/pentingnya-pembelajaran-bahasa- inggris-di-sekolah/ Ahmad Hasanudin. (2021). Peningkatan Keterampilan Berbicara dengan Metode Debat Plus pada Pembelajaran Bahasa Inggris. Jurnal Pendidikan Dan Sastra Inggris, 1(2), 19–22. https://doi.org/10.55606/jupensi.v1i2.1835 Berlianti, R. (2022). Pentingnya Kemampuan Berbahasa Inggris dalam Proses Pembelajaran. Kompasiana. https://www.kompasiana.com/ngjnkg/62d0b0f56e7f010d426ce4e3/pen tingnya-kemampuan-berbahasa-inggris-dalam-proses-pembelajaran Chotimah, C. (2022). the Use of Cake Application on Students’ Speaking Skill in English for Specific Purposes (Esp). Lingua, 18(1), 60–70. https://doi.org/10.34005/lingua.v18i01.1851 Dhanan Abimanto, I. M. (2023). Efektivitas Penggunaan Teknologi AI Dalam Pembelajaran Bahasa Inggris. Sinar Dunia: Jurnal Riset Sosial Humaniora Dan Ilmu Pendidikan, 2(2), 256–266. IDEAS, Vol. 13, No. 2, December 2025 ISSN 2338-4778 (Print) ISSN 2548-4192 (Online) 5459 Febriani, I., Sari, P., & Syahrir, S. (2023). Enhancing Students’ English Pronunciation With Cake Game Application. IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature, 11(2), 1091– 1100. https://doi.org/10.24256/ideas.v11i2.3625 Heaton, J. B. (1988). Writing English language tests. Longman. I Gede Santika Yasa, Luh Made Dwi Wedayanthi, M. D. S. (2023). Analisis Kesulitan Belajar Bahasa Inggris Siswa Kelas XII AP 1 SMK Singamandawa Tahun 2023. Ilmiah Kependidikan, 3, 5423–5434. Iman, N. (2020). 12 Aplikasi Belajar Bahasa Asing untuk Android, Makin Mahir Tanpa Kursus. DailySocial. https://dailysocial.id/post/aplikasi-belajar- bahasa-asing#:~:text=Cake merupakan salah satu aplikasi,berukuran 7%2C3 MB saja. Irma Herawati. (2023). Pemanfaatan AI dalam Pembelajaran Bahasa Inggris. Jonglojateng.https://joglojateng.com/2023/10/06/pemanfaatan-ai- dalam-pembelajaran-bahasa-inggris/ Maharani Putri. (2023). Motivasi Belajar Mahasiswa Pada Pemahaman Membaca Bahasa Inggris. Jurnal Penelitian Dan Pengkajian Ilmiah Sosial Budaya, 2(1), 31–34. https://doi.org/10.47233/jppisb.v2i1.658 Nasrum, M., Sari, P., & Nasriandi, N. (2023). Improve Students’ English Vocabulary Using the Smart TTS game App. IDEAS: Journal on English Language Teaching and Learning, Linguistics and Literature, 11(2), 1081– 1090. https://doi.org/10.24256/ideas.v11i2.3634 Noorman Haryadi, R. (2020). PENGARUH KEBIASAAN MEMBACA TERHADAP KEMAMPUAN BERBICARA BAHASA INGGRIS SMA Negeri 99 Jakarta. Jurnal Manajemen Bisnis Dan Keuangan, 1(2), 14–30. https://doi.org/10.51805/jmbk.v1i2.15 Oyebola Olusola Ayeni, Nancy Mohd Al Hamad, Onyebuchi Nneamaka Chisom, Blessing Osawaru, & Ololade Elizabeth Adewusi. (2024). AI in education: A review of personalized learning and educational technology. GSC Advanced Research and Reviews, 18(2), 261–271. https://doi.org/10.30574/gscarr.2024.18.2.0062 Qiao, H., & Zhao, A. (2023). Artificial intelligence-based language learning: illuminating the impact on speaking skills and self-regulation in Chinese EFL context. Frontiers in Psychology, 14(November). https://doi.org/10.3389/fpsyg.2023.1255594 Rahmawati, Mulyani, R. (2024). IMPROVING SPEAKING SKILLS BY USING PROJECT-BASED LEARNING WITH THE DUOLINGO APPLICATION FOR THE SECOND GRADE STUDENTS OF SMAKON ACEH. 5(1), 1–12. Saka, A. B., Oyedele, L. O., Akanbi, L. A., Ganiyu, S. A., Chan, D. W., & Bello, S. A. (2023). Conversational artificial intelligence in the AEC industry: A review of present status, challenges and opportunities. Advanced Engineering Informatics, 55, 101869. Israeni Iskandar, Nasriandi, Syahrir Artificial Intelligence -Based Language Learning: Cake Application in Developing Speaking Skill 5460 scott thornbury. (2017). Scott Thornbury and his 30 Language Teaching Methods. Cambridge.https://www.cambridge.org/elt/blog/2017/12/06/scott- thornbury-and-his-30-language-teaching-methods/ Siti Amalia Rachmawati. (2023). Pengaruh Penggunaan Aplikasi Cake Apps terhadap Kemampuan Berbicara Siswa. PROGRESSA: Journal of Islamic Religious Instruction, 7(2), 160–169. https://doi.org/10.32616/pgr.v7.2.467.160-169 Werdamurti. (2023). Peningkatan Keterampilan Berbicara Dalam Bahasa Inggris Melalui Teknik Three Steps Interview Di Kelas X Smk Negeri 4 Tanah Grogot. VOCATIONAL: Jurnal Inovasi Pendidikan Kejuruan, 3(2), 78–87. https://doi.org/10.51878/vocational.v3i2.2247 Zou, B., Du, Y., Wang, Z., Chen, J., & Zhang, W. (2023). An Investigation Into Artificial Intelligence Speech Evaluation Programs With Automatic Feedback for Developing EFL Learners’ Speaking Skills. SAGE Open, 13(3). https://doi.org/10.1177/21582440231193818