PENGENDALIAN BIT RATE BERDASARKAN METODE SILENCE SUPPRESSION PADA PENGIRIMAN DATA SUARA

• Main Author: KETUT BUDI MULIARTA, I
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/3201
• Institution: Institut Teknologi Bandung

<b>Abstract:</b><i><br>Recent improvements of multimedia data handling and computer networks have facilitate the wide use of transmission of multimedia data over computer networks, which imply that many types of data, including text, image, speech, and video, would have to share the network bandwidth. Because running out of bandwidth could lead to data loss, bandwidth conservation by decreasing the data size is commonly used, although it usually means decreasing the quality. However, if the bandwidth is big enough, we would certainly prefer to deliver data in its highest quality instead of the compressed version of it. The solution is to use a delivery method that can use the bandwidth optimally, i.e. sending data in smaller size (with lower quality) if available bandwidth is small, and sending data with higher quality (in bigger size) if available bandwidth is big enough. <p>In this thesis work, a rate control mechanism using silence suppression method with adaptive threshold determination to deliver speech over computer networks is proposed. The silence suppression method is used to decrease the data size by excluding the silence regions, which are determined using a regularly updated threshold value. This adaptive threshold determination enables the silence suppression method to give a consistent result regardless of how loud the speaker's voice is. Variation in speech loudness could cause a problem if we use a constant threshold value. The rate control mechanism monitors the network condition to make a decision that effects the threshold value computation, which in turn will play an important role in determining the data size. </p>Test results show that the rate control mechanism can facilitate the optimal use of available bandwidth. If available bandwidth is big enough, speech is sent at a higher bit rate, so speech can be received in good quality. If available bandwidth is small, speech is sent at a lower rate, ensuring that the whole speech can still be received, although in lower quality. We can improve the performance of this mechanism by using several speech formats, instead of only one, so it can deliver speech at an even lower rate with relatively good quality.