Citation Context

...using the dictionary defined in (4), we rewrite the model (3) as Note that this three-layer model is similar to the sines + transients + noise models used in many low bit-rate parametric audio coders =-=[17]-=-. In essence, in the generative model described by (5), is an error term, i.e, the noise term, that will be modeled as Gaussian white noise with variance . 2 A central ingredient of the model to be pr...

Citation Context

...ds is fully “object-based,” and this results in a limited compression gain. For instance, at 6 kbit/s, HILN performs only slightly better than a simple parametric coder [5], but not as well as TwinVQ =-=[8]-=-. In this paper, we propose a Bayesian approach to decompose music signals into pitched sound objects for very low bit-rate object coding purposes. We do not focus on accurately estimating the fundame...

Citation Context

...inusoidal quantization scheme, where quantization of frequency parameters is done independently. 1. INTRODUCTION Parametric coding is an efficient tool for representing audio signals at low bit rates =-=[1, 2]-=-. A common parametric model decomposes an audio signal into three components: a sinusoidal component, a noise component and a transient component, each of which are coded by different subcoders. The s...

Citation Context

...presentation of an audio signal for very low bite-rate compression purposes [1]. More precisely, the audio transient compact representation by parametric models is an up-to-date and difficult problem =-=[2]-=-, [3], [4]. Basically, in an audio signal, we have two important features : the spectral content and the time waveform. For some kind of quasi-stationary signals, the most important is to well represe...

Citation Context

...cies can be estimated by smoothing the output of a monophonic pitch tracker [8] but otherwise a polyphonic pitch tracker is needed [9]. The MPEG4 HILN (Harmonic and Individual Lines plus Noise) coder =-=[10]-=- combines a single predominant harmonic stack with other standalone tracks. Amplitudes within a harmonic stack are typically described by quantized MFCCs or LPCs [9, 10] at a few key time frames, othe...

Citation Context

...lity which makes model based approaches amenable to both scalability and packet loss robustness. An excellent review of recent developments in model based audio compression techniques can be found in =-=[1]-=-. The scalable compression scheme described in this paper is similar in certain respects to existing work. There are a number of audio compression schemes that use sinusoidal modeling, e.g., [2, 3, 4]...

Citation Context

...ne other possible applications of our prioritization scheme before concluding. 2. RELATED WORK While parametric, progressive and scalable codecs are a key research topic in the audio coding community =-=[1, 2, 3]-=-, few attempts to date have been made to design scalable or selective approaches for real-time signal processing. Fouad et al. [4] propose a level-of-detail rendering approach for spatialized audio wh...

Citation Context

... with a general audio coder [2]. For coding at very low bitrates, so-called parametric speech and audio coding techniques can provide better coding efficiency than the both approaches described above =-=[3, 4, 5]-=-, especially for signals with low complexity content like clean speech or single instruments. To provide optimal coding efficiency for all types of audio material over a large range of bitrates, the M...

Citation Context

...ently represented by sinusoidal modeling. Hence, a sinusoidal model is often combined with additional signal models for noise and transients. Overviews of such hybrid models can be found e.g. in [2], =-=[3]. -=-The time-discrete representation of the signal x(n) is denoted (1) x[n] = x((n + 1/2)(1/ fs)), (2) ∗ The author is now with Coding Technologies Sweden AB, Stockholm, Sweden. MPCA02-1 where fs is the...